1PERLFUNC(1)            Perl Programmers Reference Guide            PERLFUNC(1)
2
3
4

NAME

6       perlfunc - Perl builtin functions
7

DESCRIPTION

9       The functions in this section can serve as terms in an expression.
10       They fall into two major categories: list operators and named unary
11       operators.  These differ in their precedence relationship with a
12       following comma.  (See the precedence table in perlop.)  List operators
13       take more than one argument, while unary operators can never take more
14       than one argument.  Thus, a comma terminates the argument of a unary
15       operator, but merely separates the arguments of a list operator.  A
16       unary operator generally provides scalar context to its argument, while
17       a list operator may provide either scalar or list contexts for its
18       arguments.  If it does both, scalar arguments come first and list
19       argument follow, and there can only ever be one such list argument.
20       For instance, "splice" has three scalar arguments followed by a list,
21       whereas "gethostbyname" has four scalar arguments.
22
23       In the syntax descriptions that follow, list operators that expect a
24       list (and provide list context for elements of the list) are shown with
25       LIST as an argument.  Such a list may consist of any combination of
26       scalar arguments or list values; the list values will be included in
27       the list as if each individual element were interpolated at that point
28       in the list, forming a longer single-dimensional list value.  Commas
29       should separate literal elements of the LIST.
30
31       Any function in the list below may be used either with or without
32       parentheses around its arguments.  (The syntax descriptions omit the
33       parentheses.)  If you use parentheses, the simple but occasionally
34       surprising rule is this: It looks like a function, therefore it is a
35       function, and precedence doesn't matter.  Otherwise it's a list
36       operator or unary operator, and precedence does matter.  Whitespace
37       between the function and left parenthesis doesn't count, so sometimes
38       you need to be careful:
39
40           print 1+2+4;      # Prints 7.
41           print(1+2) + 4;   # Prints 3.
42           print (1+2)+4;    # Also prints 3!
43           print +(1+2)+4;   # Prints 7.
44           print ((1+2)+4);  # Prints 7.
45
46       If you run Perl with the "use warnings" pragma, it can warn you about
47       this.  For example, the third line above produces:
48
49           print (...) interpreted as function at - line 1.
50           Useless use of integer addition in void context at - line 1.
51
52       A few functions take no arguments at all, and therefore work as neither
53       unary nor list operators.  These include such functions as "time" and
54       "endpwent".  For example, "time+86_400" always means "time() + 86_400".
55
56       For functions that can be used in either a scalar or list context,
57       nonabortive failure is generally indicated in scalar context by
58       returning the undefined value, and in list context by returning the
59       empty list.
60
61       Remember the following important rule: There is no rule that relates
62       the behavior of an expression in list context to its behavior in scalar
63       context, or vice versa.  It might do two totally different things.
64       Each operator and function decides which sort of value would be most
65       appropriate to return in scalar context.  Some operators return the
66       length of the list that would have been returned in list context.  Some
67       operators return the first value in the list.  Some operators return
68       the last value in the list.  Some operators return a count of
69       successful operations.  In general, they do what you want, unless you
70       want consistency.
71
72       A named array in scalar context is quite different from what would at
73       first glance appear to be a list in scalar context.  You can't get a
74       list like "(1,2,3)" into being in scalar context, because the compiler
75       knows the context at compile time.  It would generate the scalar comma
76       operator there, not the list concatenation version of the comma.  That
77       means it was never a list to start with.
78
79       In general, functions in Perl that serve as wrappers for system calls
80       ("syscalls") of the same name (like chown(2), fork(2), closedir(2),
81       etc.) return true when they succeed and "undef" otherwise, as is
82       usually mentioned in the descriptions below.  This is different from
83       the C interfaces, which return "-1" on failure.  Exceptions to this
84       rule include "wait", "waitpid", and "syscall".  System calls also set
85       the special $! variable on failure.  Other functions do not, except
86       accidentally.
87
88       Extension modules can also hook into the Perl parser to define new
89       kinds of keyword-headed expression.  These may look like functions, but
90       may also look completely different.  The syntax following the keyword
91       is defined entirely by the extension.  If you are an implementor, see
92       "PL_keyword_plugin" in perlapi for the mechanism.  If you are using
93       such a module, see the module's documentation for details of the syntax
94       that it defines.
95
96   Perl Functions by Category
97       Here are Perl's functions (including things that look like functions,
98       like some keywords and named operators) arranged by category.  Some
99       functions appear in more than one place.  Any warnings, including those
100       produced by keywords, are described in perldiag and warnings.
101
102       Functions for SCALARs or strings
103           "chomp", "chop", "chr", "crypt", "fc", "hex", "index", "lc",
104           "lcfirst", "length", "oct", "ord", "pack", "q//", "qq//",
105           "reverse", "rindex", "sprintf", "substr", "tr///", "uc", "ucfirst",
106           "y///"
107
108           "fc" is available only if the "fc" feature is enabled or if it is
109           prefixed with "CORE::".  The "fc" feature is enabled automatically
110           with a "use v5.16" (or higher) declaration in the current scope.
111
112       Regular expressions and pattern matching
113           "m//", "pos", "qr//", "quotemeta", "s///", "split", "study"
114
115       Numeric functions
116           "abs", "atan2", "cos", "exp", "hex", "int", "log", "oct", "rand",
117           "sin", "sqrt", "srand"
118
119       Functions for real @ARRAYs
120           "each", "keys", "pop", "push", "shift", "splice", "unshift",
121           "values"
122
123       Functions for list data
124           "grep", "join", "map", "qw//", "reverse", "sort", "unpack"
125
126       Functions for real %HASHes
127           "delete", "each", "exists", "keys", "values"
128
129       Input and output functions
130           "binmode", "close", "closedir", "dbmclose", "dbmopen", "die",
131           "eof", "fileno", "flock", "format", "getc", "print", "printf",
132           "read", "readdir", "readline", "rewinddir", "say", "seek",
133           "seekdir", "select", "syscall", "sysread", "sysseek", "syswrite",
134           "tell", "telldir", "truncate", "warn", "write"
135
136           "say" is available only if the "say" feature is enabled or if it is
137           prefixed with "CORE::".  The "say" feature is enabled automatically
138           with a "use v5.10" (or higher) declaration in the current scope.
139
140       Functions for fixed-length data or records
141           "pack", "read", "syscall", "sysread", "sysseek", "syswrite",
142           "unpack", "vec"
143
144       Functions for filehandles, files, or directories
145           "-X", "chdir", "chmod", "chown", "chroot", "fcntl", "glob",
146           "ioctl", "link", "lstat", "mkdir", "open", "opendir", "readlink",
147           "rename", "rmdir", "select", "stat", "symlink", "sysopen", "umask",
148           "unlink", "utime"
149
150       Keywords related to the control flow of your Perl program
151           "break", "caller", "continue", "die", "do", "dump", "eval",
152           "evalbytes", "exit", "__FILE__", "goto", "last", "__LINE__",
153           "next", "__PACKAGE__", "redo", "return", "sub", "__SUB__",
154           "wantarray"
155
156           "break" is available only if you enable the experimental "switch"
157           feature or use the "CORE::" prefix.  The "switch" feature also
158           enables the "default", "given" and "when" statements, which are
159           documented in "Switch Statements" in perlsyn.  The "switch" feature
160           is enabled automatically with a "use v5.10" (or higher) declaration
161           in the current scope.  In Perl v5.14 and earlier, "continue"
162           required the "switch" feature, like the other keywords.
163
164           "evalbytes" is only available with the "evalbytes" feature (see
165           feature) or if prefixed with "CORE::".  "__SUB__" is only available
166           with the "current_sub" feature or if prefixed with "CORE::".  Both
167           the "evalbytes" and "current_sub" features are enabled
168           automatically with a "use v5.16" (or higher) declaration in the
169           current scope.
170
171       Keywords related to scoping
172           "caller", "import", "local", "my", "our", "package", "state", "use"
173
174           "state" is available only if the "state" feature is enabled or if
175           it is prefixed with "CORE::".  The "state" feature is enabled
176           automatically with a "use v5.10" (or higher) declaration in the
177           current scope.
178
179       Miscellaneous functions
180           "defined", "formline", "lock", "prototype", "reset", "scalar",
181           "undef"
182
183       Functions for processes and process groups
184           "alarm", "exec", "fork", "getpgrp", "getppid", "getpriority",
185           "kill", "pipe", "qx//", "readpipe", "setpgrp", "setpriority",
186           "sleep", "system", "times", "wait", "waitpid"
187
188       Keywords related to Perl modules
189           "do", "import", "no", "package", "require", "use"
190
191       Keywords related to classes and object-orientation
192           "bless", "dbmclose", "dbmopen", "package", "ref", "tie", "tied",
193           "untie", "use"
194
195       Low-level socket functions
196           "accept", "bind", "connect", "getpeername", "getsockname",
197           "getsockopt", "listen", "recv", "send", "setsockopt", "shutdown",
198           "socket", "socketpair"
199
200       System V interprocess communication functions
201           "msgctl", "msgget", "msgrcv", "msgsnd", "semctl", "semget",
202           "semop", "shmctl", "shmget", "shmread", "shmwrite"
203
204       Fetching user and group info
205           "endgrent", "endhostent", "endnetent", "endpwent", "getgrent",
206           "getgrgid", "getgrnam", "getlogin", "getpwent", "getpwnam",
207           "getpwuid", "setgrent", "setpwent"
208
209       Fetching network info
210           "endprotoent", "endservent", "gethostbyaddr", "gethostbyname",
211           "gethostent", "getnetbyaddr", "getnetbyname", "getnetent",
212           "getprotobyname", "getprotobynumber", "getprotoent",
213           "getservbyname", "getservbyport", "getservent", "sethostent",
214           "setnetent", "setprotoent", "setservent"
215
216       Time-related functions
217           "gmtime", "localtime", "time", "times"
218
219       Non-function keywords
220           "and", "AUTOLOAD", "BEGIN", "catch", "CHECK", "cmp", "CORE",
221           "__DATA__", "default", "defer", "DESTROY", "else", "elseif",
222           "elsif", "END", "__END__", "eq", "finally", "for", "foreach", "ge",
223           "given", "gt", "if", "INIT", "isa", "le", "lt", "ne", "not", "or",
224           "try", "UNITCHECK", "unless", "until", "when", "while", "x", "xor"
225
226   Portability
227       Perl was born in Unix and can therefore access all common Unix system
228       calls.  In non-Unix environments, the functionality of some Unix system
229       calls may not be available or details of the available functionality
230       may differ slightly.  The Perl functions affected by this are:
231
232       "-X", "binmode", "chmod", "chown", "chroot", "crypt", "dbmclose",
233       "dbmopen", "dump", "endgrent", "endhostent", "endnetent",
234       "endprotoent", "endpwent", "endservent", "exec", "fcntl", "flock",
235       "fork", "getgrent", "getgrgid", "gethostbyname", "gethostent",
236       "getlogin", "getnetbyaddr", "getnetbyname", "getnetent", "getppid",
237       "getpgrp", "getpriority", "getprotobynumber", "getprotoent",
238       "getpwent", "getpwnam", "getpwuid", "getservbyport", "getservent",
239       "getsockopt", "glob", "ioctl", "kill", "link", "lstat", "msgctl",
240       "msgget", "msgrcv", "msgsnd", "open", "pipe", "readlink", "rename",
241       "select", "semctl", "semget", "semop", "setgrent", "sethostent",
242       "setnetent", "setpgrp", "setpriority", "setprotoent", "setpwent",
243       "setservent", "setsockopt", "shmctl", "shmget", "shmread", "shmwrite",
244       "socket", "socketpair", "stat", "symlink", "syscall", "sysopen",
245       "system", "times", "truncate", "umask", "unlink", "utime", "wait",
246       "waitpid"
247
248       For more information about the portability of these functions, see
249       perlport and other available platform-specific documentation.
250
251   Alphabetical Listing of Perl Functions
252       -X FILEHANDLE
253       -X EXPR
254       -X DIRHANDLE
255       -X  A file test, where X is one of the letters listed below.  This
256           unary operator takes one argument, either a filename, a filehandle,
257           or a dirhandle, and tests the associated file to see if something
258           is true about it.  If the argument is omitted, tests $_, except for
259           "-t", which tests STDIN.  Unless otherwise documented, it returns 1
260           for true and '' for false.  If the file doesn't exist or can't be
261           examined, it returns "undef" and sets $! (errno).  With the
262           exception of the "-l" test they all follow symbolic links because
263           they use "stat()" and not "lstat()" (so dangling symlinks can't be
264           examined and will therefore report failure).
265
266           Despite the funny names, precedence is the same as any other named
267           unary operator.  The operator may be any of:
268
269               -r  File is readable by effective uid/gid.
270               -w  File is writable by effective uid/gid.
271               -x  File is executable by effective uid/gid.
272               -o  File is owned by effective uid.
273
274               -R  File is readable by real uid/gid.
275               -W  File is writable by real uid/gid.
276               -X  File is executable by real uid/gid.
277               -O  File is owned by real uid.
278
279               -e  File exists.
280               -z  File has zero size (is empty).
281               -s  File has nonzero size (returns size in bytes).
282
283               -f  File is a plain file.
284               -d  File is a directory.
285               -l  File is a symbolic link (false if symlinks aren't
286                   supported by the file system).
287               -p  File is a named pipe (FIFO), or Filehandle is a pipe.
288               -S  File is a socket.
289               -b  File is a block special file.
290               -c  File is a character special file.
291               -t  Filehandle is opened to a tty.
292
293               -u  File has setuid bit set.
294               -g  File has setgid bit set.
295               -k  File has sticky bit set.
296
297               -T  File is an ASCII or UTF-8 text file (heuristic guess).
298               -B  File is a "binary" file (opposite of -T).
299
300               -M  Script start time minus file modification time, in days.
301               -A  Same for access time.
302               -C  Same for inode change time (Unix, may differ for other
303                   platforms)
304
305           Example:
306
307               while (<>) {
308                   chomp;
309                   next unless -f $_;  # ignore specials
310                   #...
311               }
312
313           Note that "-s/a/b/" does not do a negated substitution.  Saying
314           "-exp($foo)" still works as expected, however: only single letters
315           following a minus are interpreted as file tests.
316
317           These operators are exempt from the "looks like a function rule"
318           described above.  That is, an opening parenthesis after the
319           operator does not affect how much of the following code constitutes
320           the argument.  Put the opening parentheses before the operator to
321           separate it from code that follows (this applies only to operators
322           with higher precedence than unary operators, of course):
323
324               -s($file) + 1024   # probably wrong; same as -s($file + 1024)
325               (-s $file) + 1024  # correct
326
327           The interpretation of the file permission operators "-r", "-R",
328           "-w", "-W", "-x", and "-X" is by default based solely on the mode
329           of the file and the uids and gids of the user.  There may be other
330           reasons you can't actually read, write, or execute the file: for
331           example network filesystem access controls, ACLs (access control
332           lists), read-only filesystems, and unrecognized executable formats.
333           Note that the use of these six specific operators to verify if some
334           operation is possible is usually a mistake, because it may be open
335           to race conditions.
336
337           Also note that, for the superuser on the local filesystems, the
338           "-r", "-R", "-w", and "-W" tests always return 1, and "-x" and "-X"
339           return 1 if any execute bit is set in the mode.  Scripts run by the
340           superuser may thus need to do a "stat" to determine the actual mode
341           of the file, or temporarily set their effective uid to something
342           else.
343
344           If you are using ACLs, there is a pragma called "filetest" that may
345           produce more accurate results than the bare "stat" mode bits.  When
346           under "use filetest 'access'", the above-mentioned filetests test
347           whether the permission can(not) be granted using the access(2)
348           family of system calls.  Also note that the "-x" and "-X" tests may
349           under this pragma return true even if there are no execute
350           permission bits set (nor any extra execute permission ACLs).  This
351           strangeness is due to the underlying system calls' definitions.
352           Note also that, due to the implementation of "use filetest
353           'access'", the "_" special filehandle won't cache the results of
354           the file tests when this pragma is in effect.  Read the
355           documentation for the "filetest" pragma for more information.
356
357           The "-T" and "-B" tests work as follows.  The first block or so of
358           the file is examined to see if it is valid UTF-8 that includes non-
359           ASCII characters.  If so, it's a "-T" file.  Otherwise, that same
360           portion of the file is examined for odd characters such as strange
361           control codes or characters with the high bit set.  If more than a
362           third of the characters are strange, it's a "-B" file; otherwise
363           it's a "-T" file.  Also, any file containing a zero byte in the
364           examined portion is considered a binary file.  (If executed within
365           the scope of a use locale which includes "LC_CTYPE", odd characters
366           are anything that isn't a printable nor space in the current
367           locale.)  If "-T" or "-B" is used on a filehandle, the current IO
368           buffer is examined rather than the first block.  Both "-T" and "-B"
369           return true on an empty file, or a file at EOF when testing a
370           filehandle.  Because you have to read a file to do the "-T" test,
371           on most occasions you want to use a "-f" against the file first, as
372           in "next unless -f $file && -T $file".
373
374           If any of the file tests (or either the "stat" or "lstat" operator)
375           is given the special filehandle consisting of a solitary underline,
376           then the stat structure of the previous file test (or "stat"
377           operator) is used, saving a system call.  (This doesn't work with
378           "-t", and you need to remember that "lstat" and "-l" leave values
379           in the stat structure for the symbolic link, not the real file.)
380           (Also, if the stat buffer was filled by an "lstat" call, "-T" and
381           "-B" will reset it with the results of "stat _").  Example:
382
383               print "Can do.\n" if -r $a || -w _ || -x _;
384
385               stat($filename);
386               print "Readable\n" if -r _;
387               print "Writable\n" if -w _;
388               print "Executable\n" if -x _;
389               print "Setuid\n" if -u _;
390               print "Setgid\n" if -g _;
391               print "Sticky\n" if -k _;
392               print "Text\n" if -T _;
393               print "Binary\n" if -B _;
394
395           As of Perl 5.10.0, as a form of purely syntactic sugar, you can
396           stack file test operators, in a way that "-f -w -x $file" is
397           equivalent to "-x $file && -w _ && -f _".  (This is only fancy
398           syntax: if you use the return value of "-f $file" as an argument to
399           another filetest operator, no special magic will happen.)
400
401           Portability issues: "-X" in perlport.
402
403           To avoid confusing would-be users of your code with mysterious
404           syntax errors, put something like this at the top of your script:
405
406               use v5.10;  # so filetest ops can stack
407
408       abs VALUE
409       abs Returns the absolute value of its argument.  If VALUE is omitted,
410           uses $_.
411
412       accept NEWSOCKET,GENERICSOCKET
413           Accepts an incoming socket connect, just as accept(2) does.
414           Returns the packed address if it succeeded, false otherwise.  See
415           the example in "Sockets: Client/Server Communication" in perlipc.
416
417           On systems that support a close-on-exec flag on files, the flag
418           will be set for the newly opened file descriptor, as determined by
419           the value of $^F.  See "$^F" in perlvar.
420
421       alarm SECONDS
422       alarm
423           Arranges to have a SIGALRM delivered to this process after the
424           specified number of wallclock seconds has elapsed.  If SECONDS is
425           not specified, the value stored in $_ is used.  (On some machines,
426           unfortunately, the elapsed time may be up to one second less or
427           more than you specified because of how seconds are counted, and
428           process scheduling may delay the delivery of the signal even
429           further.)
430
431           Only one timer may be counting at once.  Each call disables the
432           previous timer, and an argument of 0 may be supplied to cancel the
433           previous timer without starting a new one.  The returned value is
434           the amount of time remaining on the previous timer.
435
436           For delays of finer granularity than one second, the Time::HiRes
437           module (from CPAN, and starting from Perl 5.8 part of the standard
438           distribution) provides "ualarm".  You may also use Perl's four-
439           argument version of "select" leaving the first three arguments
440           undefined, or you might be able to use the "syscall" interface to
441           access setitimer(2) if your system supports it.  See perlfaq8 for
442           details.
443
444           It is usually a mistake to intermix "alarm" and "sleep" calls,
445           because "sleep" may be internally implemented on your system with
446           "alarm".
447
448           If you want to use "alarm" to time out a system call you need to
449           use an "eval"/"die" pair.  You can't rely on the alarm causing the
450           system call to fail with $! set to "EINTR" because Perl sets up
451           signal handlers to restart system calls on some systems.  Using
452           "eval"/"die" always works, modulo the caveats given in "Signals" in
453           perlipc.
454
455               eval {
456                   local $SIG{ALRM} = sub { die "alarm\n" }; # NB: \n required
457                   alarm $timeout;
458                   my $nread = sysread $socket, $buffer, $size;
459                   alarm 0;
460               };
461               if ($@) {
462                   die unless $@ eq "alarm\n";   # propagate unexpected errors
463                   # timed out
464               }
465               else {
466                   # didn't
467               }
468
469           For more information see perlipc.
470
471           Portability issues: "alarm" in perlport.
472
473       atan2 Y,X
474           Returns the arctangent of Y/X in the range -PI to PI.
475
476           For the tangent operation, you may use the "Math::Trig::tan"
477           function, or use the familiar relation:
478
479               sub tan { sin($_[0]) / cos($_[0])  }
480
481           The return value for "atan2(0,0)" is implementation-defined;
482           consult your atan2(3) manpage for more information.
483
484           Portability issues: "atan2" in perlport.
485
486       bind SOCKET,NAME
487           Binds a network address to a socket, just as bind(2) does.  Returns
488           true if it succeeded, false otherwise.  NAME should be a packed
489           address of the appropriate type for the socket.  See the examples
490           in "Sockets: Client/Server Communication" in perlipc.
491
492       binmode FILEHANDLE, LAYER
493       binmode FILEHANDLE
494           Arranges for FILEHANDLE to be read or written in "binary" or "text"
495           mode on systems where the run-time libraries distinguish between
496           binary and text files.  If FILEHANDLE is an expression, the value
497           is taken as the name of the filehandle.  Returns true on success,
498           otherwise it returns "undef" and sets $! (errno).
499
500           On some systems (in general, DOS- and Windows-based systems)
501           "binmode" is necessary when you're not working with a text file.
502           For the sake of portability it is a good idea always to use it when
503           appropriate, and never to use it when it isn't appropriate.  Also,
504           people can set their I/O to be by default UTF8-encoded Unicode, not
505           bytes.
506
507           In other words: regardless of platform, use "binmode" on binary
508           data, like images, for example.
509
510           If LAYER is present it is a single string, but may contain multiple
511           directives.  The directives alter the behaviour of the filehandle.
512           When LAYER is present, using binmode on a text file makes sense.
513
514           If LAYER is omitted or specified as ":raw" the filehandle is made
515           suitable for passing binary data.  This includes turning off
516           possible CRLF translation and marking it as bytes (as opposed to
517           Unicode characters).  Note that, despite what may be implied in
518           "Programming Perl" (the Camel, 3rd edition) or elsewhere, ":raw" is
519           not simply the inverse of ":crlf".  Other layers that would affect
520           the binary nature of the stream are also disabled.  See PerlIO, and
521           the discussion about the PERLIO environment variable in perlrun.
522
523           The ":bytes", ":crlf", ":utf8", and any other directives of the
524           form ":...", are called I/O layers.  The open pragma can be used to
525           establish default I/O layers.
526
527           The LAYER parameter of the "binmode" function is described as
528           "DISCIPLINE" in "Programming Perl, 3rd Edition".  However, since
529           the publishing of this book, by many known as "Camel III", the
530           consensus of the naming of this functionality has moved from
531           "discipline" to "layer".  All documentation of this version of Perl
532           therefore refers to "layers" rather than to "disciplines".  Now
533           back to the regularly scheduled documentation...
534
535           To mark FILEHANDLE as UTF-8, use ":utf8" or ":encoding(UTF-8)".
536           ":utf8" just marks the data as UTF-8 without further checking,
537           while ":encoding(UTF-8)" checks the data for actually being valid
538           UTF-8.  More details can be found in PerlIO::encoding.
539
540           In general, "binmode" should be called after "open" but before any
541           I/O is done on the filehandle.  Calling "binmode" normally flushes
542           any pending buffered output data (and perhaps pending input data)
543           on the handle.  An exception to this is the ":encoding" layer that
544           changes the default character encoding of the handle.  The
545           ":encoding" layer sometimes needs to be called in mid-stream, and
546           it doesn't flush the stream.  ":encoding" also implicitly pushes on
547           top of itself the ":utf8" layer because internally Perl operates on
548           UTF8-encoded Unicode characters.
549
550           The operating system, device drivers, C libraries, and Perl run-
551           time system all conspire to let the programmer treat a single
552           character ("\n") as the line terminator, irrespective of external
553           representation.  On many operating systems, the native text file
554           representation matches the internal representation, but on some
555           platforms the external representation of "\n" is made up of more
556           than one character.
557
558           All variants of Unix, Mac OS (old and new), and Stream_LF files on
559           VMS use a single character to end each line in the external
560           representation of text (even though that single character is
561           CARRIAGE RETURN on old, pre-Darwin flavors of Mac OS, and is LINE
562           FEED on Unix and most VMS files).  In other systems like OS/2, DOS,
563           and the various flavors of MS-Windows, your program sees a "\n" as
564           a simple "\cJ", but what's stored in text files are the two
565           characters "\cM\cJ".  That means that if you don't use "binmode" on
566           these systems, "\cM\cJ" sequences on disk will be converted to "\n"
567           on input, and any "\n" in your program will be converted back to
568           "\cM\cJ" on output.  This is what you want for text files, but it
569           can be disastrous for binary files.
570
571           Another consequence of using "binmode" (on some systems) is that
572           special end-of-file markers will be seen as part of the data
573           stream.  For systems from the Microsoft family this means that, if
574           your binary data contain "\cZ", the I/O subsystem will regard it as
575           the end of the file, unless you use "binmode".
576
577           "binmode" is important not only for "readline" and "print"
578           operations, but also when using "read", "seek", "sysread",
579           "syswrite" and "tell" (see perlport for more details).  See the $/
580           and "$\" variables in perlvar for how to manually set your input
581           and output line-termination sequences.
582
583           Portability issues: "binmode" in perlport.
584
585       bless REF,CLASSNAME
586       bless REF
587           This function tells the thingy referenced by REF that it is now an
588           object in the CLASSNAME package.  If CLASSNAME is an empty string,
589           it is interpreted as referring to the "main" package.  If CLASSNAME
590           is omitted, the current package is used.  Because a "bless" is
591           often the last thing in a constructor, it returns the reference for
592           convenience.  Always use the two-argument version if a derived
593           class might inherit the method doing the blessing.  See perlobj for
594           more about the blessing (and blessings) of objects.
595
596           Consider always blessing objects in CLASSNAMEs that are mixed case.
597           Namespaces with all lowercase names are considered reserved for
598           Perl pragmas.  Builtin types have all uppercase names.  To prevent
599           confusion, you may wish to avoid such package names as well.  It is
600           advised to avoid the class name 0, because much code erroneously
601           uses the result of "ref" as a truth value.
602
603           See "Perl Modules" in perlmod.
604
605       break
606           Break out of a "given" block.
607
608           "break" is available only if the "switch" feature is enabled or if
609           it is prefixed with "CORE::". The "switch" feature is enabled
610           automatically with a "use v5.10" (or higher) declaration in the
611           current scope.
612
613       caller EXPR
614       caller
615           Returns the context of the current pure perl subroutine call.  In
616           scalar context, returns the caller's package name if there is a
617           caller (that is, if we're in a subroutine or "eval" or "require")
618           and the undefined value otherwise.  caller never returns XS subs
619           and they are skipped.  The next pure perl sub will appear instead
620           of the XS sub in caller's return values.  In list context, caller
621           returns
622
623                  # 0         1          2
624               my ($package, $filename, $line) = caller;
625
626           Like "__FILE__" and "__LINE__", the filename and line number
627           returned here may be altered by the mechanism described at "Plain
628           Old Comments (Not!)" in perlsyn.
629
630           With EXPR, it returns some extra information that the debugger uses
631           to print a stack trace.  The value of EXPR indicates how many call
632           frames to go back before the current one.
633
634               #  0         1          2      3            4
635            my ($package, $filename, $line, $subroutine, $hasargs,
636
637               #  5          6          7            8       9         10
638               $wantarray, $evaltext, $is_require, $hints, $bitmask, $hinthash)
639             = caller($i);
640
641           Here, $subroutine is the function that the caller called (rather
642           than the function containing the caller).  Note that $subroutine
643           may be "(eval)" if the frame is not a subroutine call, but an
644           "eval".  In such a case additional elements $evaltext and
645           $is_require are set: $is_require is true if the frame is created by
646           a "require" or "use" statement, $evaltext contains the text of the
647           "eval EXPR" statement.  In particular, for an "eval BLOCK"
648           statement, $subroutine is "(eval)", but $evaltext is undefined.
649           (Note also that each "use" statement creates a "require" frame
650           inside an "eval EXPR" frame.)  $subroutine may also be "(unknown)"
651           if this particular subroutine happens to have been deleted from the
652           symbol table.  $hasargs is true if a new instance of @_ was set up
653           for the frame.  $hints and $bitmask contain pragmatic hints that
654           the caller was compiled with.  $hints corresponds to $^H, and
655           $bitmask corresponds to "${^WARNING_BITS}".  The $hints and
656           $bitmask values are subject to change between versions of Perl, and
657           are not meant for external use.
658
659           $hinthash is a reference to a hash containing the value of "%^H"
660           when the caller was compiled, or "undef" if "%^H" was empty.  Do
661           not modify the values of this hash, as they are the actual values
662           stored in the optree.
663
664           Note that the only types of call frames that are visible are
665           subroutine calls and "eval". Other forms of context, such as
666           "while" or "foreach" loops or "try" blocks are not considered
667           interesting to "caller", as they do not alter the behaviour of the
668           "return" expression.
669
670           Furthermore, when called from within the DB package in list
671           context, and with an argument, caller returns more detailed
672           information: it sets the list variable @DB::args to be the
673           arguments with which the subroutine was invoked.
674
675           Be aware that the optimizer might have optimized call frames away
676           before "caller" had a chance to get the information.  That means
677           that caller(N) might not return information about the call frame
678           you expect it to, for "N > 1".  In particular, @DB::args might have
679           information from the previous time "caller" was called.
680
681           Be aware that setting @DB::args is best effort, intended for
682           debugging or generating backtraces, and should not be relied upon.
683           In particular, as @_ contains aliases to the caller's arguments,
684           Perl does not take a copy of @_, so @DB::args will contain
685           modifications the subroutine makes to @_ or its contents, not the
686           original values at call time.  @DB::args, like @_, does not hold
687           explicit references to its elements, so under certain cases its
688           elements may have become freed and reallocated for other variables
689           or temporary values.  Finally, a side effect of the current
690           implementation is that the effects of "shift @_" can normally be
691           undone (but not "pop @_" or other splicing, and not if a reference
692           to @_ has been taken, and subject to the caveat about reallocated
693           elements), so @DB::args is actually a hybrid of the current state
694           and initial state of @_.  Buyer beware.
695
696       chdir EXPR
697       chdir FILEHANDLE
698       chdir DIRHANDLE
699       chdir
700           Changes the working directory to EXPR, if possible.  If EXPR is
701           omitted, changes to the directory specified by $ENV{HOME}, if set;
702           if not, changes to the directory specified by $ENV{LOGDIR}.  (Under
703           VMS, the variable $ENV{'SYS$LOGIN'} is also checked, and used if it
704           is set.)  If neither is set, "chdir" does nothing and fails.  It
705           returns true on success, false otherwise.  See the example under
706           "die".
707
708           On systems that support fchdir(2), you may pass a filehandle or
709           directory handle as the argument.  On systems that don't support
710           fchdir(2), passing handles raises an exception.
711
712       chmod LIST
713           Changes the permissions of a list of files.  The first element of
714           the list must be the numeric mode, which should probably be an
715           octal number, and which definitely should not be a string of octal
716           digits: 0644 is okay, but "0644" is not.  Returns the number of
717           files successfully changed.  See also "oct" if all you have is a
718           string.
719
720               my $cnt = chmod 0755, "foo", "bar";
721               chmod 0755, @executables;
722               my $mode = "0644"; chmod $mode, "foo";      # !!! sets mode to
723                                                           # --w----r-T
724               my $mode = "0644"; chmod oct($mode), "foo"; # this is better
725               my $mode = 0644;   chmod $mode, "foo";      # this is best
726
727           On systems that support fchmod(2), you may pass filehandles among
728           the files.  On systems that don't support fchmod(2), passing
729           filehandles raises an exception.  Filehandles must be passed as
730           globs or glob references to be recognized; barewords are considered
731           filenames.
732
733               open(my $fh, "<", "foo");
734               my $perm = (stat $fh)[2] & 07777;
735               chmod($perm | 0600, $fh);
736
737           You can also import the symbolic "S_I*" constants from the "Fcntl"
738           module:
739
740               use Fcntl qw( :mode );
741               chmod S_IRWXU|S_IRGRP|S_IXGRP|S_IROTH|S_IXOTH, @executables;
742               # Identical to the chmod 0755 of the example above.
743
744           Portability issues: "chmod" in perlport.
745
746       chomp VARIABLE
747       chomp( LIST )
748       chomp
749           This safer version of "chop" removes any trailing string that
750           corresponds to the current value of $/ (also known as
751           $INPUT_RECORD_SEPARATOR in the "English" module).  It returns the
752           total number of characters removed from all its arguments.  It's
753           often used to remove the newline from the end of an input record
754           when you're worried that the final record may be missing its
755           newline.  When in paragraph mode ("$/ = ''"), it removes all
756           trailing newlines from the string.  When in slurp mode ("$/ =
757           undef") or fixed-length record mode ($/ is a reference to an
758           integer or the like; see perlvar), "chomp" won't remove anything.
759           If VARIABLE is omitted, it chomps $_.  Example:
760
761               while (<>) {
762                   chomp;  # avoid \n on last field
763                   my @array = split(/:/);
764                   # ...
765               }
766
767           If VARIABLE is a hash, it chomps the hash's values, but not its
768           keys, resetting the "each" iterator in the process.
769
770           You can actually chomp anything that's an lvalue, including an
771           assignment:
772
773               chomp(my $cwd = `pwd`);
774               chomp(my $answer = <STDIN>);
775
776           If you chomp a list, each element is chomped, and the total number
777           of characters removed is returned.
778
779           Note that parentheses are necessary when you're chomping anything
780           that is not a simple variable.  This is because "chomp $cwd =
781           `pwd`;" is interpreted as "(chomp $cwd) = `pwd`;", rather than as
782           "chomp( $cwd = `pwd` )" which you might expect.  Similarly, "chomp
783           $a, $b" is interpreted as "chomp($a), $b" rather than as "chomp($a,
784           $b)".
785
786       chop VARIABLE
787       chop( LIST )
788       chop
789           Chops off the last character of a string and returns the character
790           chopped.  It is much more efficient than "s/.$//s" because it
791           neither scans nor copies the string.  If VARIABLE is omitted, chops
792           $_.  If VARIABLE is a hash, it chops the hash's values, but not its
793           keys, resetting the "each" iterator in the process.
794
795           You can actually chop anything that's an lvalue, including an
796           assignment.
797
798           If you chop a list, each element is chopped.  Only the value of the
799           last "chop" is returned.
800
801           Note that "chop" returns the last character.  To return all but the
802           last character, use "substr($string, 0, -1)".
803
804           See also "chomp".
805
806       chown LIST
807           Changes the owner (and group) of a list of files.  The first two
808           elements of the list must be the numeric uid and gid, in that
809           order.  A value of -1 in either position is interpreted by most
810           systems to leave that value unchanged.  Returns the number of files
811           successfully changed.
812
813               my $cnt = chown $uid, $gid, 'foo', 'bar';
814               chown $uid, $gid, @filenames;
815
816           On systems that support fchown(2), you may pass filehandles among
817           the files.  On systems that don't support fchown(2), passing
818           filehandles raises an exception.  Filehandles must be passed as
819           globs or glob references to be recognized; barewords are considered
820           filenames.
821
822           Here's an example that looks up nonnumeric uids in the passwd file:
823
824               print "User: ";
825               chomp(my $user = <STDIN>);
826               print "Files: ";
827               chomp(my $pattern = <STDIN>);
828
829               my ($login,$pass,$uid,$gid) = getpwnam($user)
830                   or die "$user not in passwd file";
831
832               my @ary = glob($pattern);  # expand filenames
833               chown $uid, $gid, @ary;
834
835           On most systems, you are not allowed to change the ownership of the
836           file unless you're the superuser, although you should be able to
837           change the group to any of your secondary groups.  On insecure
838           systems, these restrictions may be relaxed, but this is not a
839           portable assumption.  On POSIX systems, you can detect this
840           condition this way:
841
842               use POSIX qw(sysconf _PC_CHOWN_RESTRICTED);
843               my $can_chown_giveaway = ! sysconf(_PC_CHOWN_RESTRICTED);
844
845           Portability issues: "chown" in perlport.
846
847       chr NUMBER
848       chr Returns the character represented by that NUMBER in the character
849           set.  For example, "chr(65)" is "A" in either ASCII or Unicode, and
850           chr(0x263a) is a Unicode smiley face.
851
852           Negative values give the Unicode replacement character
853           (chr(0xfffd)), except under the bytes pragma, where the low eight
854           bits of the value (truncated to an integer) are used.
855
856           If NUMBER is omitted, uses $_.
857
858           For the reverse, use "ord".
859
860           Note that characters from 128 to 255 (inclusive) are by default
861           internally not encoded as UTF-8 for backward compatibility reasons.
862
863           See perlunicode for more about Unicode.
864
865       chroot FILENAME
866       chroot
867           This function works like the system call by the same name: it makes
868           the named directory the new root directory for all further
869           pathnames that begin with a "/" by your process and all its
870           children.  (It doesn't change your current working directory, which
871           is unaffected.)  For security reasons, this call is restricted to
872           the superuser.  If FILENAME is omitted, does a "chroot" to $_.
873
874           NOTE:  It is mandatory for security to "chdir("/")" ("chdir" to the
875           root directory) immediately after a "chroot", otherwise the current
876           working directory may be outside of the new root.
877
878           Portability issues: "chroot" in perlport.
879
880       close FILEHANDLE
881       close
882           Closes the file or pipe associated with the filehandle, flushes the
883           IO buffers, and closes the system file descriptor.  Returns true if
884           those operations succeed and if no error was reported by any PerlIO
885           layer.  Closes the currently selected filehandle if the argument is
886           omitted.
887
888           You don't have to close FILEHANDLE if you are immediately going to
889           do another "open" on it, because "open" closes it for you.  (See
890           "open".) However, an explicit "close" on an input file resets the
891           line counter ($.), while the implicit close done by "open" does
892           not.
893
894           If the filehandle came from a piped open, "close" returns false if
895           one of the other syscalls involved fails or if its program exits
896           with non-zero status.  If the only problem was that the program
897           exited non-zero, $! will be set to 0.  Closing a pipe also waits
898           for the process executing on the pipe to exit--in case you wish to
899           look at the output of the pipe afterwards--and implicitly puts the
900           exit status value of that command into $? and
901           "${^CHILD_ERROR_NATIVE}".
902
903           If there are multiple threads running, "close" on a filehandle from
904           a piped open returns true without waiting for the child process to
905           terminate, if the filehandle is still open in another thread.
906
907           Closing the read end of a pipe before the process writing to it at
908           the other end is done writing results in the writer receiving a
909           SIGPIPE.  If the other end can't handle that, be sure to read all
910           the data before closing the pipe.
911
912           Example:
913
914               open(OUTPUT, '|sort >foo')  # pipe to sort
915                   or die "Can't start sort: $!";
916               #...                        # print stuff to output
917               close OUTPUT                # wait for sort to finish
918                   or warn $! ? "Error closing sort pipe: $!"
919                              : "Exit status $? from sort";
920               open(INPUT, 'foo')          # get sort's results
921                   or die "Can't open 'foo' for input: $!";
922
923           FILEHANDLE may be an expression whose value can be used as an
924           indirect filehandle, usually the real filehandle name or an
925           autovivified handle.
926
927       closedir DIRHANDLE
928           Closes a directory opened by "opendir" and returns the success of
929           that system call.
930
931       connect SOCKET,NAME
932           Attempts to connect to a remote socket, just like connect(2).
933           Returns true if it succeeded, false otherwise.  NAME should be a
934           packed address of the appropriate type for the socket.  See the
935           examples in "Sockets: Client/Server Communication" in perlipc.
936
937       continue BLOCK
938       continue
939           When followed by a BLOCK, "continue" is actually a flow control
940           statement rather than a function.  If there is a "continue" BLOCK
941           attached to a BLOCK (typically in a "while" or "foreach"), it is
942           always executed just before the conditional is about to be
943           evaluated again, just like the third part of a "for" loop in C.
944           Thus it can be used to increment a loop variable, even when the
945           loop has been continued via the "next" statement (which is similar
946           to the C "continue" statement).
947
948           "last", "next", or "redo" may appear within a "continue" block;
949           "last" and "redo" behave as if they had been executed within the
950           main block.  So will "next", but since it will execute a "continue"
951           block, it may be more entertaining.
952
953               while (EXPR) {
954                   ### redo always comes here
955                   do_something;
956               } continue {
957                   ### next always comes here
958                   do_something_else;
959                   # then back the top to re-check EXPR
960               }
961               ### last always comes here
962
963           Omitting the "continue" section is equivalent to using an empty
964           one, logically enough, so "next" goes directly back to check the
965           condition at the top of the loop.
966
967           When there is no BLOCK, "continue" is a function that falls through
968           the current "when" or "default" block instead of iterating a
969           dynamically enclosing "foreach" or exiting a lexically enclosing
970           "given".  In Perl 5.14 and earlier, this form of "continue" was
971           only available when the "switch" feature was enabled.  See feature
972           and "Switch Statements" in perlsyn for more information.
973
974       cos EXPR
975       cos Returns the cosine of EXPR (expressed in radians).  If EXPR is
976           omitted, takes the cosine of $_.
977
978           For the inverse cosine operation, you may use the
979           "Math::Trig::acos" function, or use this relation:
980
981               sub acos { atan2( sqrt(1 - $_[0] * $_[0]), $_[0] ) }
982
983       crypt PLAINTEXT,SALT
984           Creates a digest string exactly like the crypt(3) function in the C
985           library (assuming that you actually have a version there that has
986           not been extirpated as a potential munition).
987
988           "crypt" is a one-way hash function.  The PLAINTEXT and SALT are
989           turned into a short string, called a digest, which is returned.
990           The same PLAINTEXT and SALT will always return the same string, but
991           there is no (known) way to get the original PLAINTEXT from the
992           hash.  Small changes in the PLAINTEXT or SALT will result in large
993           changes in the digest.
994
995           There is no decrypt function.  This function isn't all that useful
996           for cryptography (for that, look for Crypt modules on your nearby
997           CPAN mirror) and the name "crypt" is a bit of a misnomer.  Instead
998           it is primarily used to check if two pieces of text are the same
999           without having to transmit or store the text itself.  An example is
1000           checking if a correct password is given.  The digest of the
1001           password is stored, not the password itself.  The user types in a
1002           password that is "crypt"'d with the same salt as the stored digest.
1003           If the two digests match, the password is correct.
1004
1005           When verifying an existing digest string you should use the digest
1006           as the salt (like "crypt($plain, $digest) eq $digest").  The SALT
1007           used to create the digest is visible as part of the digest.  This
1008           ensures "crypt" will hash the new string with the same salt as the
1009           digest.  This allows your code to work with the standard "crypt"
1010           and with more exotic implementations.  In other words, assume
1011           nothing about the returned string itself nor about how many bytes
1012           of SALT may matter.
1013
1014           Traditionally the result is a string of 13 bytes: two first bytes
1015           of the salt, followed by 11 bytes from the set "[./0-9A-Za-z]", and
1016           only the first eight bytes of PLAINTEXT mattered.  But alternative
1017           hashing schemes (like MD5), higher level security schemes (like
1018           C2), and implementations on non-Unix platforms may produce
1019           different strings.
1020
1021           When choosing a new salt create a random two character string whose
1022           characters come from the set "[./0-9A-Za-z]" (like "join '', ('.',
1023           '/', 0..9, 'A'..'Z', 'a'..'z')[rand 64, rand 64]").  This set of
1024           characters is just a recommendation; the characters allowed in the
1025           salt depend solely on your system's crypt library, and Perl can't
1026           restrict what salts "crypt" accepts.
1027
1028           Here's an example that makes sure that whoever runs this program
1029           knows their password:
1030
1031               my $pwd = (getpwuid($<))[1];
1032
1033               system "stty -echo";
1034               print "Password: ";
1035               chomp(my $word = <STDIN>);
1036               print "\n";
1037               system "stty echo";
1038
1039               if (crypt($word, $pwd) ne $pwd) {
1040                   die "Sorry...\n";
1041               } else {
1042                   print "ok\n";
1043               }
1044
1045           Of course, typing in your own password to whoever asks you for it
1046           is unwise.
1047
1048           The "crypt" function is unsuitable for hashing large quantities of
1049           data, not least of all because you can't get the information back.
1050           Look at the Digest module for more robust algorithms.
1051
1052           If using "crypt" on a Unicode string (which potentially has
1053           characters with codepoints above 255), Perl tries to make sense of
1054           the situation by trying to downgrade (a copy of) the string back to
1055           an eight-bit byte string before calling "crypt" (on that copy).  If
1056           that works, good.  If not, "crypt" dies with "Wide character in
1057           crypt".
1058
1059           Portability issues: "crypt" in perlport.
1060
1061       dbmclose HASH
1062           [This function has been largely superseded by the "untie"
1063           function.]
1064
1065           Breaks the binding between a DBM file and a hash.
1066
1067           Portability issues: "dbmclose" in perlport.
1068
1069       dbmopen HASH,DBNAME,MASK
1070           [This function has been largely superseded by the "tie" function.]
1071
1072           This binds a dbm(3), ndbm(3), sdbm(3), gdbm(3), or Berkeley DB file
1073           to a hash.  HASH is the name of the hash.  (Unlike normal "open",
1074           the first argument is not a filehandle, even though it looks like
1075           one).  DBNAME is the name of the database (without the .dir or .pag
1076           extension if any).  If the database does not exist, it is created
1077           with protection specified by MASK (as modified by the "umask").  To
1078           prevent creation of the database if it doesn't exist, you may
1079           specify a MODE of 0, and the function will return a false value if
1080           it can't find an existing database.  If your system supports only
1081           the older DBM functions, you may make only one "dbmopen" call in
1082           your program.  In older versions of Perl, if your system had
1083           neither DBM nor ndbm, calling "dbmopen" produced a fatal error; it
1084           now falls back to sdbm(3).
1085
1086           If you don't have write access to the DBM file, you can only read
1087           hash variables, not set them.  If you want to test whether you can
1088           write, either use file tests or try setting a dummy hash entry
1089           inside an "eval" to trap the error.
1090
1091           Note that functions such as "keys" and "values" may return huge
1092           lists when used on large DBM files.  You may prefer to use the
1093           "each" function to iterate over large DBM files.  Example:
1094
1095               # print out history file offsets
1096               dbmopen(%HIST,'/usr/lib/news/history',0666);
1097               while (($key,$val) = each %HIST) {
1098                   print $key, ' = ', unpack('L',$val), "\n";
1099               }
1100               dbmclose(%HIST);
1101
1102           See also AnyDBM_File for a more general description of the pros and
1103           cons of the various dbm approaches, as well as DB_File for a
1104           particularly rich implementation.
1105
1106           You can control which DBM library you use by loading that library
1107           before you call "dbmopen":
1108
1109               use DB_File;
1110               dbmopen(%NS_Hist, "$ENV{HOME}/.netscape/history.db")
1111                   or die "Can't open netscape history file: $!";
1112
1113           Portability issues: "dbmopen" in perlport.
1114
1115       defined EXPR
1116       defined
1117           Returns a Boolean value telling whether EXPR has a value other than
1118           the undefined value "undef".  If EXPR is not present, $_ is
1119           checked.
1120
1121           Many operations return "undef" to indicate failure, end of file,
1122           system error, uninitialized variable, and other exceptional
1123           conditions.  This function allows you to distinguish "undef" from
1124           other values.  (A simple Boolean test will not distinguish among
1125           "undef", zero, the empty string, and "0", which are all equally
1126           false.)  Note that since "undef" is a valid scalar, its presence
1127           doesn't necessarily indicate an exceptional condition: "pop"
1128           returns "undef" when its argument is an empty array, or when the
1129           element to return happens to be "undef".
1130
1131           You may also use "defined(&func)" to check whether subroutine
1132           "func" has ever been defined.  The return value is unaffected by
1133           any forward declarations of "func".  A subroutine that is not
1134           defined may still be callable: its package may have an "AUTOLOAD"
1135           method that makes it spring into existence the first time that it
1136           is called; see perlsub.
1137
1138           Use of "defined" on aggregates (hashes and arrays) is no longer
1139           supported. It used to report whether memory for that aggregate had
1140           ever been allocated.  You should instead use a simple test for
1141           size:
1142
1143               if (@an_array) { print "has array elements\n" }
1144               if (%a_hash)   { print "has hash members\n"   }
1145
1146           When used on a hash element, it tells you whether the value is
1147           defined, not whether the key exists in the hash.  Use "exists" for
1148           the latter purpose.
1149
1150           Examples:
1151
1152               print if defined $switch{D};
1153               print "$val\n" while defined($val = pop(@ary));
1154               die "Can't readlink $sym: $!"
1155                   unless defined($value = readlink $sym);
1156               sub foo { defined &$bar ? $bar->(@_) : die "No bar"; }
1157               $debugging = 0 unless defined $debugging;
1158
1159           Note:  Many folks tend to overuse "defined" and are then surprised
1160           to discover that the number 0 and "" (the zero-length string) are,
1161           in fact, defined values.  For example, if you say
1162
1163               "ab" =~ /a(.*)b/;
1164
1165           The pattern match succeeds and $1 is defined, although it matched
1166           "nothing".  It didn't really fail to match anything.  Rather, it
1167           matched something that happened to be zero characters long.  This
1168           is all very above-board and honest.  When a function returns an
1169           undefined value, it's an admission that it couldn't give you an
1170           honest answer.  So you should use "defined" only when questioning
1171           the integrity of what you're trying to do.  At other times, a
1172           simple comparison to 0 or "" is what you want.
1173
1174           See also "undef", "exists", "ref".
1175
1176       delete EXPR
1177           Given an expression that specifies an element or slice of a hash,
1178           "delete" deletes the specified elements from that hash so that
1179           "exists" on that element no longer returns true.  Setting a hash
1180           element to the undefined value does not remove its key, but
1181           deleting it does; see "exists".
1182
1183           In list context, usually returns the value or values deleted, or
1184           the last such element in scalar context.  The return list's length
1185           corresponds to that of the argument list: deleting non-existent
1186           elements returns the undefined value in their corresponding
1187           positions. Since Perl 5.28, a key/value hash slice can be passed to
1188           "delete", and the return value is a list of key/value pairs (two
1189           elements for each item deleted from the hash).
1190
1191           "delete" may also be used on arrays and array slices, but its
1192           behavior is less straightforward.  Although "exists" will return
1193           false for deleted entries, deleting array elements never changes
1194           indices of existing values; use "shift" or "splice" for that.
1195           However, if any deleted elements fall at the end of an array, the
1196           array's size shrinks to the position of the highest element that
1197           still tests true for "exists", or to 0 if none do.  In other words,
1198           an array won't have trailing nonexistent elements after a delete.
1199
1200           WARNING: Calling "delete" on array values is strongly discouraged.
1201           The notion of deleting or checking the existence of Perl array
1202           elements is not conceptually coherent, and can lead to surprising
1203           behavior.
1204
1205           Deleting from %ENV modifies the environment.  Deleting from a hash
1206           tied to a DBM file deletes the entry from the DBM file.  Deleting
1207           from a "tied" hash or array may not necessarily return anything; it
1208           depends on the implementation of the "tied" package's DELETE
1209           method, which may do whatever it pleases.
1210
1211           The "delete local EXPR" construct localizes the deletion to the
1212           current block at run time.  Until the block exits, elements locally
1213           deleted temporarily no longer exist.  See "Localized deletion of
1214           elements of composite types" in perlsub.
1215
1216               my %hash = (foo => 11, bar => 22, baz => 33);
1217               my $scalar = delete $hash{foo};         # $scalar is 11
1218               $scalar = delete @hash{qw(foo bar)}; # $scalar is 22
1219               my @array  = delete @hash{qw(foo baz)}; # @array  is (undef,33)
1220
1221           The following (inefficiently) deletes all the values of %HASH and
1222           @ARRAY:
1223
1224               foreach my $key (keys %HASH) {
1225                   delete $HASH{$key};
1226               }
1227
1228               foreach my $index (0 .. $#ARRAY) {
1229                   delete $ARRAY[$index];
1230               }
1231
1232           And so do these:
1233
1234               delete @HASH{keys %HASH};
1235
1236               delete @ARRAY[0 .. $#ARRAY];
1237
1238           But both are slower than assigning the empty list or undefining
1239           %HASH or @ARRAY, which is the customary way to empty out an
1240           aggregate:
1241
1242               %HASH = ();     # completely empty %HASH
1243               undef %HASH;    # forget %HASH ever existed
1244
1245               @ARRAY = ();    # completely empty @ARRAY
1246               undef @ARRAY;   # forget @ARRAY ever existed
1247
1248           The EXPR can be arbitrarily complicated provided its final
1249           operation is an element or slice of an aggregate:
1250
1251               delete $ref->[$x][$y]{$key};
1252               delete $ref->[$x][$y]->@{$key1, $key2, @morekeys};
1253
1254               delete $ref->[$x][$y][$index];
1255               delete $ref->[$x][$y]->@[$index1, $index2, @moreindices];
1256
1257       die LIST
1258           "die" raises an exception.  Inside an "eval" the exception is
1259           stuffed into $@ and the "eval" is terminated with the undefined
1260           value.  If the exception is outside of all enclosing "eval"s, then
1261           the uncaught exception is printed to "STDERR" and perl exits with
1262           an exit code indicating failure.  If you need to exit the process
1263           with a specific exit code, see "exit".
1264
1265           Equivalent examples:
1266
1267               die "Can't cd to spool: $!\n" unless chdir '/usr/spool/news';
1268               chdir '/usr/spool/news' or die "Can't cd to spool: $!\n"
1269
1270           Most of the time, "die" is called with a string to use as the
1271           exception.  You may either give a single non-reference operand to
1272           serve as the exception, or a list of two or more items, which will
1273           be stringified and concatenated to make the exception.
1274
1275           If the string exception does not end in a newline, the current
1276           script line number and input line number (if any) and a newline are
1277           appended to it.  Note that the "input line number" (also known as
1278           "chunk") is subject to whatever notion of "line" happens to be
1279           currently in effect, and is also available as the special variable
1280           $..  See "$/" in perlvar and "$." in perlvar.
1281
1282           Hint: sometimes appending ", stopped" to your message will cause it
1283           to make better sense when the string "at foo line 123" is appended.
1284           Suppose you are running script "canasta".
1285
1286               die "/etc/games is no good";
1287               die "/etc/games is no good, stopped";
1288
1289           produce, respectively
1290
1291               /etc/games is no good at canasta line 123.
1292               /etc/games is no good, stopped at canasta line 123.
1293
1294           If LIST was empty or made an empty string, and $@ already contains
1295           an exception value (typically from a previous "eval"), then that
1296           value is reused after appending "\t...propagated".  This is useful
1297           for propagating exceptions:
1298
1299               eval { ... };
1300               die unless $@ =~ /Expected exception/;
1301
1302           If LIST was empty or made an empty string, and $@ contains an
1303           object reference that has a "PROPAGATE" method, that method will be
1304           called with additional file and line number parameters.  The return
1305           value replaces the value in $@;  i.e., as if "$@ = eval {
1306           $@->PROPAGATE(__FILE__, __LINE__) };" were called.
1307
1308           If LIST was empty or made an empty string, and $@ is also empty,
1309           then the string "Died" is used.
1310
1311           You can also call "die" with a reference argument, and if this is
1312           trapped within an "eval", $@ contains that reference.  This permits
1313           more elaborate exception handling using objects that maintain
1314           arbitrary state about the exception.  Such a scheme is sometimes
1315           preferable to matching particular string values of $@ with regular
1316           expressions.
1317
1318           Because Perl stringifies uncaught exception messages before
1319           display, you'll probably want to overload stringification
1320           operations on exception objects.  See overload for details about
1321           that.  The stringified message should be non-empty, and should end
1322           in a newline, in order to fit in with the treatment of string
1323           exceptions.  Also, because an exception object reference cannot be
1324           stringified without destroying it, Perl doesn't attempt to append
1325           location or other information to a reference exception.  If you
1326           want location information with a complex exception object, you'll
1327           have to arrange to put the location information into the object
1328           yourself.
1329
1330           Because $@ is a global variable, be careful that analyzing an
1331           exception caught by "eval" doesn't replace the reference in the
1332           global variable.  It's easiest to make a local copy of the
1333           reference before any manipulations.  Here's an example:
1334
1335               use Scalar::Util "blessed";
1336
1337               eval { ... ; die Some::Module::Exception->new( FOO => "bar" ) };
1338               if (my $ev_err = $@) {
1339                   if (blessed($ev_err)
1340                       && $ev_err->isa("Some::Module::Exception")) {
1341                       # handle Some::Module::Exception
1342                   }
1343                   else {
1344                       # handle all other possible exceptions
1345                   }
1346               }
1347
1348           If an uncaught exception results in interpreter exit, the exit code
1349           is determined from the values of $! and $? with this pseudocode:
1350
1351               exit $! if $!;              # errno
1352               exit $? >> 8 if $? >> 8;    # child exit status
1353               exit 255;                   # last resort
1354
1355           As with "exit", $? is set prior to unwinding the call stack; any
1356           "DESTROY" or "END" handlers can then alter this value, and thus
1357           Perl's exit code.
1358
1359           The intent is to squeeze as much possible information about the
1360           likely cause into the limited space of the system exit code.
1361           However, as $! is the value of C's "errno", which can be set by any
1362           system call, this means that the value of the exit code used by
1363           "die" can be non-predictable, so should not be relied upon, other
1364           than to be non-zero.
1365
1366           You can arrange for a callback to be run just before the "die" does
1367           its deed, by setting the $SIG{__DIE__} hook.  The associated
1368           handler is called with the exception as an argument, and can change
1369           the exception, if it sees fit, by calling "die" again.  See "%SIG"
1370           in perlvar for details on setting %SIG entries, and "eval" for some
1371           examples.  Although this feature was to be run only right before
1372           your program was to exit, this is not currently so: the
1373           $SIG{__DIE__} hook is currently called even inside "eval"ed
1374           blocks/strings!  If one wants the hook to do nothing in such
1375           situations, put
1376
1377               die @_ if $^S;
1378
1379           as the first line of the handler (see "$^S" in perlvar).  Because
1380           this promotes strange action at a distance, this counterintuitive
1381           behavior may be fixed in a future release.
1382
1383           See also "exit", "warn", and the Carp module.
1384
1385       do BLOCK
1386           Not really a function.  Returns the value of the last command in
1387           the sequence of commands indicated by BLOCK.  When modified by the
1388           "while" or "until" loop modifier, executes the BLOCK once before
1389           testing the loop condition.  (On other statements the loop
1390           modifiers test the conditional first.)
1391
1392           "do BLOCK" does not count as a loop, so the loop control statements
1393           "next", "last", or "redo" cannot be used to leave or restart the
1394           block.  See perlsyn for alternative strategies.
1395
1396       do EXPR
1397           Uses the value of EXPR as a filename and executes the contents of
1398           the file as a Perl script:
1399
1400               # load the exact specified file (./ and ../ special-cased)
1401               do '/foo/stat.pl';
1402               do './stat.pl';
1403               do '../foo/stat.pl';
1404
1405               # search for the named file within @INC
1406               do 'stat.pl';
1407               do 'foo/stat.pl';
1408
1409           "do './stat.pl'" is largely like
1410
1411               eval `cat stat.pl`;
1412
1413           except that it's more concise, runs no external processes, and
1414           keeps track of the current filename for error messages. It also
1415           differs in that code evaluated with "do FILE" cannot see lexicals
1416           in the enclosing scope; "eval STRING" does.  It's the same,
1417           however, in that it does reparse the file every time you call it,
1418           so you probably don't want to do this inside a loop.
1419
1420           Using "do" with a relative path (except for ./ and ../), like
1421
1422               do 'foo/stat.pl';
1423
1424           will search the @INC directories, and update %INC if the file is
1425           found.  See "@INC" in perlvar and "%INC" in perlvar for these
1426           variables. In particular, note that whilst historically @INC
1427           contained '.' (the current directory) making these two cases
1428           equivalent, that is no longer necessarily the case, as '.' is not
1429           included in @INC by default in perl versions 5.26.0 onwards.
1430           Instead, perl will now warn:
1431
1432               do "stat.pl" failed, '.' is no longer in @INC;
1433               did you mean do "./stat.pl"?
1434
1435           If "do" can read the file but cannot compile it, it returns "undef"
1436           and sets an error message in $@.  If "do" cannot read the file, it
1437           returns undef and sets $! to the error.  Always check $@ first, as
1438           compilation could fail in a way that also sets $!.  If the file is
1439           successfully compiled, "do" returns the value of the last
1440           expression evaluated.
1441
1442           Inclusion of library modules is better done with the "use" and
1443           "require" operators, which also do automatic error checking and
1444           raise an exception if there's a problem.
1445
1446           You might like to use "do" to read in a program configuration file.
1447           Manual error checking can be done this way:
1448
1449               # Read in config files: system first, then user.
1450               # Beware of using relative pathnames here.
1451               for $file ("/share/prog/defaults.rc",
1452                          "$ENV{HOME}/.someprogrc")
1453               {
1454                   unless ($return = do $file) {
1455                       warn "couldn't parse $file: $@" if $@;
1456                       warn "couldn't do $file: $!"    unless defined $return;
1457                       warn "couldn't run $file"       unless $return;
1458                   }
1459               }
1460
1461       dump LABEL
1462       dump EXPR
1463       dump
1464           This function causes an immediate core dump.  See also the -u
1465           command-line switch in perlrun, which does the same thing.
1466           Primarily this is so that you can use the undump program (not
1467           supplied) to turn your core dump into an executable binary after
1468           having initialized all your variables at the beginning of the
1469           program.  When the new binary is executed it will begin by
1470           executing a "goto LABEL" (with all the restrictions that "goto"
1471           suffers).  Think of it as a goto with an intervening core dump and
1472           reincarnation.  If "LABEL" is omitted, restarts the program from
1473           the top.  The "dump EXPR" form, available starting in Perl 5.18.0,
1474           allows a name to be computed at run time, being otherwise identical
1475           to "dump LABEL".
1476
1477           WARNING: Any files opened at the time of the dump will not be open
1478           any more when the program is reincarnated, with possible resulting
1479           confusion by Perl.
1480
1481           This function is now largely obsolete, mostly because it's very
1482           hard to convert a core file into an executable.  As of Perl 5.30,
1483           it must be invoked as "CORE::dump()".
1484
1485           Unlike most named operators, this has the same precedence as
1486           assignment.  It is also exempt from the looks-like-a-function rule,
1487           so "dump ("foo")."bar"" will cause "bar" to be part of the argument
1488           to "dump".
1489
1490           Portability issues: "dump" in perlport.
1491
1492       each HASH
1493       each ARRAY
1494           When called on a hash in list context, returns a 2-element list
1495           consisting of the key and value for the next element of a hash.  In
1496           Perl 5.12 and later only, it will also return the index and value
1497           for the next element of an array so that you can iterate over it;
1498           older Perls consider this a syntax error.  When called in scalar
1499           context, returns only the key (not the value) in a hash, or the
1500           index in an array.
1501
1502           Hash entries are returned in an apparently random order.  The
1503           actual random order is specific to a given hash; the exact same
1504           series of operations on two hashes may result in a different order
1505           for each hash.  Any insertion into the hash may change the order,
1506           as will any deletion, with the exception that the most recent key
1507           returned by "each" or "keys" may be deleted without changing the
1508           order.  So long as a given hash is unmodified you may rely on
1509           "keys", "values" and "each" to repeatedly return the same order as
1510           each other.  See "Algorithmic Complexity Attacks" in perlsec for
1511           details on why hash order is randomized.  Aside from the guarantees
1512           provided here the exact details of Perl's hash algorithm and the
1513           hash traversal order are subject to change in any release of Perl.
1514
1515           After "each" has returned all entries from the hash or array, the
1516           next call to "each" returns the empty list in list context and
1517           "undef" in scalar context; the next call following that one
1518           restarts iteration.  Each hash or array has its own internal
1519           iterator, accessed by "each", "keys", and "values".  The iterator
1520           is implicitly reset when "each" has reached the end as just
1521           described; it can be explicitly reset by calling "keys" or "values"
1522           on the hash or array, or by referencing the hash (but not array) in
1523           list context.  If you add or delete a hash's elements while
1524           iterating over it, the effect on the iterator is unspecified; for
1525           example, entries may be skipped or duplicated--so don't do that.
1526           Exception: It is always safe to delete the item most recently
1527           returned by "each", so the following code works properly:
1528
1529               while (my ($key, $value) = each %hash) {
1530                   print $key, "\n";
1531                   delete $hash{$key};   # This is safe
1532               }
1533
1534           Tied hashes may have a different ordering behaviour to perl's hash
1535           implementation.
1536
1537           The iterator used by "each" is attached to the hash or array, and
1538           is shared between all iteration operations applied to the same hash
1539           or array.  Thus all uses of "each" on a single hash or array
1540           advance the same iterator location.  All uses of "each" are also
1541           subject to having the iterator reset by any use of "keys" or
1542           "values" on the same hash or array, or by the hash (but not array)
1543           being referenced in list context.  This makes "each"-based loops
1544           quite fragile: it is easy to arrive at such a loop with the
1545           iterator already part way through the object, or to accidentally
1546           clobber the iterator state during execution of the loop body.  It's
1547           easy enough to explicitly reset the iterator before starting a
1548           loop, but there is no way to insulate the iterator state used by a
1549           loop from the iterator state used by anything else that might
1550           execute during the loop body.  To avoid these problems, use a
1551           "foreach" loop rather than "while"-"each".
1552
1553           This extends to using "each" on the result of an anonymous hash or
1554           array constructor.  A new underlying array or hash is created each
1555           time so each will always start iterating from scratch, eg:
1556
1557             # loops forever
1558             while (my ($key, $value) = each @{ +{ a => 1 } }) {
1559                 print "$key=$value\n";
1560             }
1561
1562           This prints out your environment like the printenv(1) program, but
1563           in a different order:
1564
1565               while (my ($key,$value) = each %ENV) {
1566                   print "$key=$value\n";
1567               }
1568
1569           Starting with Perl 5.14, an experimental feature allowed "each" to
1570           take a scalar expression. This experiment has been deemed
1571           unsuccessful, and was removed as of Perl 5.24.
1572
1573           As of Perl 5.18 you can use a bare "each" in a "while" loop, which
1574           will set $_ on every iteration.  If either an "each" expression or
1575           an explicit assignment of an "each" expression to a scalar is used
1576           as a "while"/"for" condition, then the condition actually tests for
1577           definedness of the expression's value, not for its regular truth
1578           value.
1579
1580               while (each %ENV) {
1581                   print "$_=$ENV{$_}\n";
1582               }
1583
1584           To avoid confusing would-be users of your code who are running
1585           earlier versions of Perl with mysterious syntax errors, put this
1586           sort of thing at the top of your file to signal that your code will
1587           work only on Perls of a recent vintage:
1588
1589               use v5.12;  # so keys/values/each work on arrays
1590               use v5.18;  # so each assigns to $_ in a lone while test
1591
1592           See also "keys", "values", and "sort".
1593
1594       eof FILEHANDLE
1595       eof ()
1596       eof Returns 1 if the next read on FILEHANDLE will return end of file or
1597           if FILEHANDLE is not open.  FILEHANDLE may be an expression whose
1598           value gives the real filehandle.  (Note that this function actually
1599           reads a character and then "ungetc"s it, so isn't useful in an
1600           interactive context.)  Do not read from a terminal file (or call
1601           "eof(FILEHANDLE)" on it) after end-of-file is reached.  File types
1602           such as terminals may lose the end-of-file condition if you do.
1603
1604           An "eof" without an argument uses the last file read.  Using
1605           "eof()" with empty parentheses is different.  It refers to the
1606           pseudo file formed from the files listed on the command line and
1607           accessed via the "<>" operator.  Since "<>" isn't explicitly
1608           opened, as a normal filehandle is, an "eof()" before "<>" has been
1609           used will cause @ARGV to be examined to determine if input is
1610           available.   Similarly, an "eof()" after "<>" has returned end-of-
1611           file will assume you are processing another @ARGV list, and if you
1612           haven't set @ARGV, will read input from "STDIN"; see "I/O
1613           Operators" in perlop.
1614
1615           In a "while (<>)" loop, "eof" or "eof(ARGV)" can be used to detect
1616           the end of each file, whereas "eof()" will detect the end of the
1617           very last file only.  Examples:
1618
1619               # reset line numbering on each input file
1620               while (<>) {
1621                   next if /^\s*#/;  # skip comments
1622                   print "$.\t$_";
1623               } continue {
1624                   close ARGV if eof;  # Not eof()!
1625               }
1626
1627               # insert dashes just before last line of last file
1628               while (<>) {
1629                   if (eof()) {  # check for end of last file
1630                       print "--------------\n";
1631                   }
1632                   print;
1633                   last if eof();     # needed if we're reading from a terminal
1634               }
1635
1636           Practical hint: you almost never need to use "eof" in Perl, because
1637           the input operators typically return "undef" when they run out of
1638           data or encounter an error.
1639
1640       eval EXPR
1641       eval BLOCK
1642       eval
1643           "eval" in all its forms is used to execute a little Perl program,
1644           trapping any errors encountered so they don't crash the calling
1645           program.
1646
1647           Plain "eval" with no argument is just "eval EXPR", where the
1648           expression is understood to be contained in $_.  Thus there are
1649           only two real "eval" forms; the one with an EXPR is often called
1650           "string eval".  In a string eval, the value of the expression
1651           (which is itself determined within scalar context) is first parsed,
1652           and if there were no errors, executed as a block within the lexical
1653           context of the current Perl program.  This form is typically used
1654           to delay parsing and subsequent execution of the text of EXPR until
1655           run time.  Note that the value is parsed every time the "eval"
1656           executes.
1657
1658           The other form is called "block eval".  It is less general than
1659           string eval, but the code within the BLOCK is parsed only once (at
1660           the same time the code surrounding the "eval" itself was parsed)
1661           and executed within the context of the current Perl program.  This
1662           form is typically used to trap exceptions more efficiently than the
1663           first, while also providing the benefit of checking the code within
1664           BLOCK at compile time.  BLOCK is parsed and compiled just once.
1665           Since errors are trapped, it often is used to check if a given
1666           feature is available.
1667
1668           In both forms, the value returned is the value of the last
1669           expression evaluated inside the mini-program; a return statement
1670           may also be used, just as with subroutines.  The expression
1671           providing the return value is evaluated in void, scalar, or list
1672           context, depending on the context of the "eval" itself.  See
1673           "wantarray" for more on how the evaluation context can be
1674           determined.
1675
1676           If there is a syntax error or runtime error, or a "die" statement
1677           is executed, "eval" returns "undef" in scalar context, or an empty
1678           list in list context, and $@ is set to the error message.  (Prior
1679           to 5.16, a bug caused "undef" to be returned in list context for
1680           syntax errors, but not for runtime errors.) If there was no error,
1681           $@ is set to the empty string.  A control flow operator like "last"
1682           or "goto" can bypass the setting of $@.  Beware that using "eval"
1683           neither silences Perl from printing warnings to STDERR, nor does it
1684           stuff the text of warning messages into $@.  To do either of those,
1685           you have to use the $SIG{__WARN__} facility, or turn off warnings
1686           inside the BLOCK or EXPR using "no warnings 'all'".  See "warn",
1687           perlvar, and warnings.
1688
1689           Note that, because "eval" traps otherwise-fatal errors, it is
1690           useful for determining whether a particular feature (such as
1691           "socket" or "symlink") is implemented.  It is also Perl's
1692           exception-trapping mechanism, where the "die" operator is used to
1693           raise exceptions.
1694
1695           Before Perl 5.14, the assignment to $@ occurred before restoration
1696           of localized variables, which means that for your code to run on
1697           older versions, a temporary is required if you want to mask some,
1698           but not all errors:
1699
1700            # alter $@ on nefarious repugnancy only
1701            {
1702               my $e;
1703               {
1704                 local $@; # protect existing $@
1705                 eval { test_repugnancy() };
1706                 # $@ =~ /nefarious/ and die $@; # Perl 5.14 and higher only
1707                 $@ =~ /nefarious/ and $e = $@;
1708               }
1709               die $e if defined $e
1710            }
1711
1712           There are some different considerations for each form:
1713
1714           String eval
1715               Since the return value of EXPR is executed as a block within
1716               the lexical context of the current Perl program, any outer
1717               lexical variables are visible to it, and any package variable
1718               settings or subroutine and format definitions remain
1719               afterwards.
1720
1721               Under the "unicode_eval" feature
1722                   If this feature is enabled (which is the default under a
1723                   "use 5.16" or higher declaration), Perl assumes that EXPR
1724                   is a character string.  Any "use utf8" or "no utf8"
1725                   declarations within the string thus have no effect. Source
1726                   filters are forbidden as well.  ("unicode_strings",
1727                   however, can appear within the string.)
1728
1729                   See also the "evalbytes" operator, which works properly
1730                   with source filters.
1731
1732               Outside the "unicode_eval" feature
1733                   In this case, the behavior is problematic and is not so
1734                   easily described.  Here are two bugs that cannot easily be
1735                   fixed without breaking existing programs:
1736
1737                   •   Perl's internal storage of EXPR affects the behavior of
1738                       the executed code.  For example:
1739
1740                           my $v = eval "use utf8; '$expr'";
1741
1742                       If $expr is "\xc4\x80" (U+0100 in UTF-8), then the
1743                       value stored in $v will depend on whether Perl stores
1744                       $expr "upgraded" (cf. utf8) or not:
1745
1746                       •   If upgraded, $v will be "\xc4\x80" (i.e., the "use
1747                           utf8" has no effect.)
1748
1749                       •   If non-upgraded, $v will be "\x{100}".
1750
1751                       This is undesirable since being upgraded or not should
1752                       not affect a string's behavior.
1753
1754                   •   Source filters activated within "eval" leak out into
1755                       whichever file scope is currently being compiled.  To
1756                       give an example with the CPAN module Semi::Semicolons:
1757
1758                        BEGIN { eval "use Semi::Semicolons; # not filtered" }
1759                        # filtered here!
1760
1761                       "evalbytes" fixes that to work the way one would
1762                       expect:
1763
1764                        use feature "evalbytes";
1765                        BEGIN { evalbytes "use Semi::Semicolons; # filtered" }
1766                        # not filtered
1767
1768               Problems can arise if the string expands a scalar containing a
1769               floating point number.  That scalar can expand to letters, such
1770               as "NaN" or "Infinity"; or, within the scope of a "use locale",
1771               the decimal point character may be something other than a dot
1772               (such as a comma).  None of these are likely to parse as you
1773               are likely expecting.
1774
1775               You should be especially careful to remember what's being
1776               looked at when:
1777
1778                   eval $x;        # CASE 1
1779                   eval "$x";      # CASE 2
1780
1781                   eval '$x';      # CASE 3
1782                   eval { $x };    # CASE 4
1783
1784                   eval "\$$x++";  # CASE 5
1785                   $$x++;          # CASE 6
1786
1787               Cases 1 and 2 above behave identically: they run the code
1788               contained in the variable $x.  (Although case 2 has misleading
1789               double quotes making the reader wonder what else might be
1790               happening (nothing is).)  Cases 3 and 4 likewise behave in the
1791               same way: they run the code '$x', which does nothing but return
1792               the value of $x.  (Case 4 is preferred for purely visual
1793               reasons, but it also has the advantage of compiling at compile-
1794               time instead of at run-time.)  Case 5 is a place where normally
1795               you would like to use double quotes, except that in this
1796               particular situation, you can just use symbolic references
1797               instead, as in case 6.
1798
1799               An "eval ''" executed within a subroutine defined in the "DB"
1800               package doesn't see the usual surrounding lexical scope, but
1801               rather the scope of the first non-DB piece of code that called
1802               it.  You don't normally need to worry about this unless you are
1803               writing a Perl debugger.
1804
1805               The final semicolon, if any, may be omitted from the value of
1806               EXPR.
1807
1808           Block eval
1809               If the code to be executed doesn't vary, you may use the eval-
1810               BLOCK form to trap run-time errors without incurring the
1811               penalty of recompiling each time.  The error, if any, is still
1812               returned in $@.  Examples:
1813
1814                   # make divide-by-zero nonfatal
1815                   eval { $answer = $a / $b; }; warn $@ if $@;
1816
1817                   # same thing, but less efficient
1818                   eval '$answer = $a / $b'; warn $@ if $@;
1819
1820                   # a compile-time error
1821                   eval { $answer = }; # WRONG
1822
1823                   # a run-time error
1824                   eval '$answer =';   # sets $@
1825
1826               If you want to trap errors when loading an XS module, some
1827               problems with the binary interface (such as Perl version skew)
1828               may be fatal even with "eval" unless $ENV{PERL_DL_NONLAZY} is
1829               set.  See perlrun.
1830
1831               Using the "eval {}" form as an exception trap in libraries does
1832               have some issues.  Due to the current arguably broken state of
1833               "__DIE__" hooks, you may wish not to trigger any "__DIE__"
1834               hooks that user code may have installed.  You can use the
1835               "local $SIG{__DIE__}" construct for this purpose, as this
1836               example shows:
1837
1838                   # a private exception trap for divide-by-zero
1839                   eval { local $SIG{'__DIE__'}; $answer = $a / $b; };
1840                   warn $@ if $@;
1841
1842               This is especially significant, given that "__DIE__" hooks can
1843               call "die" again, which has the effect of changing their error
1844               messages:
1845
1846                   # __DIE__ hooks may modify error messages
1847                   {
1848                      local $SIG{'__DIE__'} =
1849                             sub { (my $x = $_[0]) =~ s/foo/bar/g; die $x };
1850                      eval { die "foo lives here" };
1851                      print $@ if $@;                # prints "bar lives here"
1852                   }
1853
1854               Because this promotes action at a distance, this
1855               counterintuitive behavior may be fixed in a future release.
1856
1857               "eval BLOCK" does not count as a loop, so the loop control
1858               statements "next", "last", or "redo" cannot be used to leave or
1859               restart the block.
1860
1861               The final semicolon, if any, may be omitted from within the
1862               BLOCK.
1863
1864       evalbytes EXPR
1865       evalbytes
1866           This function is similar to a string eval, except it always parses
1867           its argument (or $_ if EXPR is omitted) as a byte string. If the
1868           string contains any code points above 255, then it cannot be a byte
1869           string, and the "evalbytes" will fail with the error stored in $@.
1870
1871           "use utf8" and "no utf8" within the string have their usual effect.
1872
1873           Source filters activated within the evaluated code apply to the
1874           code itself.
1875
1876           "evalbytes" is available starting in Perl v5.16.  To access it, you
1877           must say "CORE::evalbytes", but you can omit the "CORE::" if the
1878           "evalbytes" feature is enabled.  This is enabled automatically with
1879           a "use v5.16" (or higher) declaration in the current scope.
1880
1881       exec LIST
1882       exec PROGRAM LIST
1883           The "exec" function executes a system command and never returns;
1884           use "system" instead of "exec" if you want it to return.  It fails
1885           and returns false only if the command does not exist and it is
1886           executed directly instead of via your system's command shell (see
1887           below).
1888
1889           Since it's a common mistake to use "exec" instead of "system", Perl
1890           warns you if "exec" is called in void context and if there is a
1891           following statement that isn't "die", "warn", or "exit" (if
1892           warnings are enabled--but you always do that, right?).  If you
1893           really want to follow an "exec" with some other statement, you can
1894           use one of these styles to avoid the warning:
1895
1896               exec ('foo')   or print STDERR "couldn't exec foo: $!";
1897               { exec ('foo') }; print STDERR "couldn't exec foo: $!";
1898
1899           If there is more than one argument in LIST, this calls execvp(3)
1900           with the arguments in LIST.  If there is only one element in LIST,
1901           the argument is checked for shell metacharacters, and if there are
1902           any, the entire argument is passed to the system's command shell
1903           for parsing (this is "/bin/sh -c" on Unix platforms, but varies on
1904           other platforms).  If there are no shell metacharacters in the
1905           argument, it is split into words and passed directly to "execvp",
1906           which is more efficient.  Examples:
1907
1908               exec '/bin/echo', 'Your arguments are: ', @ARGV;
1909               exec "sort $outfile | uniq";
1910
1911           If you don't really want to execute the first argument, but want to
1912           lie to the program you are executing about its own name, you can
1913           specify the program you actually want to run as an "indirect
1914           object" (without a comma) in front of the LIST, as in "exec PROGRAM
1915           LIST".  (This always forces interpretation of the LIST as a
1916           multivalued list, even if there is only a single scalar in the
1917           list.)  Example:
1918
1919               my $shell = '/bin/csh';
1920               exec $shell '-sh';    # pretend it's a login shell
1921
1922           or, more directly,
1923
1924               exec {'/bin/csh'} '-sh';  # pretend it's a login shell
1925
1926           When the arguments get executed via the system shell, results are
1927           subject to its quirks and capabilities.  See "`STRING`" in perlop
1928           for details.
1929
1930           Using an indirect object with "exec" or "system" is also more
1931           secure.  This usage (which also works fine with "system") forces
1932           interpretation of the arguments as a multivalued list, even if the
1933           list had just one argument.  That way you're safe from the shell
1934           expanding wildcards or splitting up words with whitespace in them.
1935
1936               my @args = ( "echo surprise" );
1937
1938               exec @args;               # subject to shell escapes
1939                                           # if @args == 1
1940               exec { $args[0] } @args;  # safe even with one-arg list
1941
1942           The first version, the one without the indirect object, ran the
1943           echo program, passing it "surprise" an argument.  The second
1944           version didn't; it tried to run a program named "echo surprise",
1945           didn't find it, and set $? to a non-zero value indicating failure.
1946
1947           On Windows, only the "exec PROGRAM LIST" indirect object syntax
1948           will reliably avoid using the shell; "exec LIST", even with more
1949           than one element, will fall back to the shell if the first spawn
1950           fails.
1951
1952           Perl attempts to flush all files opened for output before the exec,
1953           but this may not be supported on some platforms (see perlport).  To
1954           be safe, you may need to set $| ($AUTOFLUSH in English) or call the
1955           "autoflush" method of "IO::Handle" on any open handles to avoid
1956           lost output.
1957
1958           Note that "exec" will not call your "END" blocks, nor will it
1959           invoke "DESTROY" methods on your objects.
1960
1961           Portability issues: "exec" in perlport.
1962
1963       exists EXPR
1964           Given an expression that specifies an element of a hash, returns
1965           true if the specified element in the hash has ever been
1966           initialized, even if the corresponding value is undefined.
1967
1968               print "Exists\n"    if exists $hash{$key};
1969               print "Defined\n"   if defined $hash{$key};
1970               print "True\n"      if $hash{$key};
1971
1972           exists may also be called on array elements, but its behavior is
1973           much less obvious and is strongly tied to the use of "delete" on
1974           arrays.
1975
1976           WARNING: Calling "exists" on array values is strongly discouraged.
1977           The notion of deleting or checking the existence of Perl array
1978           elements is not conceptually coherent, and can lead to surprising
1979           behavior.
1980
1981               print "Exists\n"    if exists $array[$index];
1982               print "Defined\n"   if defined $array[$index];
1983               print "True\n"      if $array[$index];
1984
1985           A hash or array element can be true only if it's defined and
1986           defined only if it exists, but the reverse doesn't necessarily hold
1987           true.
1988
1989           Given an expression that specifies the name of a subroutine,
1990           returns true if the specified subroutine has ever been declared,
1991           even if it is undefined.  Mentioning a subroutine name for exists
1992           or defined does not count as declaring it.  Note that a subroutine
1993           that does not exist may still be callable: its package may have an
1994           "AUTOLOAD" method that makes it spring into existence the first
1995           time that it is called; see perlsub.
1996
1997               print "Exists\n"  if exists &subroutine;
1998               print "Defined\n" if defined &subroutine;
1999
2000           Note that the EXPR can be arbitrarily complicated as long as the
2001           final operation is a hash or array key lookup or subroutine name:
2002
2003               if (exists $ref->{A}->{B}->{$key})  { }
2004               if (exists $hash{A}{B}{$key})       { }
2005
2006               if (exists $ref->{A}->{B}->[$ix])   { }
2007               if (exists $hash{A}{B}[$ix])        { }
2008
2009               if (exists &{$ref->{A}{B}{$key}})   { }
2010
2011           Although the most deeply nested array or hash element will not
2012           spring into existence just because its existence was tested, any
2013           intervening ones will.  Thus "$ref->{"A"}" and "$ref->{"A"}->{"B"}"
2014           will spring into existence due to the existence test for the $key
2015           element above.  This happens anywhere the arrow operator is used,
2016           including even here:
2017
2018               undef $ref;
2019               if (exists $ref->{"Some key"})    { }
2020               print $ref;  # prints HASH(0x80d3d5c)
2021
2022           Use of a subroutine call, rather than a subroutine name, as an
2023           argument to "exists" is an error.
2024
2025               exists &sub;    # OK
2026               exists &sub();  # Error
2027
2028       exit EXPR
2029       exit
2030           Evaluates EXPR and exits immediately with that value.    Example:
2031
2032               my $ans = <STDIN>;
2033               exit 0 if $ans =~ /^[Xx]/;
2034
2035           See also "die".  If EXPR is omitted, exits with 0 status.  The only
2036           universally recognized values for EXPR are 0 for success and 1 for
2037           error; other values are subject to interpretation depending on the
2038           environment in which the Perl program is running.  For example,
2039           exiting 69 (EX_UNAVAILABLE) from a sendmail incoming-mail filter
2040           will cause the mailer to return the item undelivered, but that's
2041           not true everywhere.
2042
2043           Don't use "exit" to abort a subroutine if there's any chance that
2044           someone might want to trap whatever error happened.  Use "die"
2045           instead, which can be trapped by an "eval".
2046
2047           The "exit" function does not always exit immediately.  It calls any
2048           defined "END" routines first, but these "END" routines may not
2049           themselves abort the exit.  Likewise any object destructors that
2050           need to be called are called before the real exit.  "END" routines
2051           and destructors can change the exit status by modifying $?.  If
2052           this is a problem, you can call "POSIX::_exit($status)" to avoid
2053           "END" and destructor processing.  See perlmod for details.
2054
2055           Portability issues: "exit" in perlport.
2056
2057       exp EXPR
2058       exp Returns e (the natural logarithm base) to the power of EXPR.  If
2059           EXPR is omitted, gives "exp($_)".
2060
2061       fc EXPR
2062       fc  Returns the casefolded version of EXPR.  This is the internal
2063           function implementing the "\F" escape in double-quoted strings.
2064
2065           Casefolding is the process of mapping strings to a form where case
2066           differences are erased; comparing two strings in their casefolded
2067           form is effectively a way of asking if two strings are equal,
2068           regardless of case.
2069
2070           Roughly, if you ever found yourself writing this
2071
2072               lc($this) eq lc($that)    # Wrong!
2073                   # or
2074               uc($this) eq uc($that)    # Also wrong!
2075                   # or
2076               $this =~ /^\Q$that\E\z/i  # Right!
2077
2078           Now you can write
2079
2080               fc($this) eq fc($that)
2081
2082           And get the correct results.
2083
2084           Perl only implements the full form of casefolding, but you can
2085           access the simple folds using "casefold()" in Unicode::UCD and
2086           "prop_invmap()" in Unicode::UCD.  For further information on
2087           casefolding, refer to the Unicode Standard, specifically sections
2088           3.13 "Default Case Operations", 4.2 "Case-Normative", and 5.18
2089           "Case Mappings", available at
2090           <https://www.unicode.org/versions/latest/>, as well as the Case
2091           Charts available at <https://www.unicode.org/charts/case/>.
2092
2093           If EXPR is omitted, uses $_.
2094
2095           This function behaves the same way under various pragmas, such as
2096           within "use feature 'unicode_strings", as "lc" does, with the
2097           single exception of "fc" of LATIN CAPITAL LETTER SHARP S (U+1E9E)
2098           within the scope of "use locale".  The foldcase of this character
2099           would normally be "ss", but as explained in the "lc" section, case
2100           changes that cross the 255/256 boundary are problematic under
2101           locales, and are hence prohibited.  Therefore, this function under
2102           locale returns instead the string "\x{17F}\x{17F}", which is the
2103           LATIN SMALL LETTER LONG S.  Since that character itself folds to
2104           "s", the string of two of them together should be equivalent to a
2105           single U+1E9E when foldcased.
2106
2107           While the Unicode Standard defines two additional forms of
2108           casefolding, one for Turkic languages and one that never maps one
2109           character into multiple characters, these are not provided by the
2110           Perl core.  However, the CPAN module "Unicode::Casing" may be used
2111           to provide an implementation.
2112
2113           "fc" is available only if the "fc" feature is enabled or if it is
2114           prefixed with "CORE::".  The "fc" feature is enabled automatically
2115           with a "use v5.16" (or higher) declaration in the current scope.
2116
2117       fcntl FILEHANDLE,FUNCTION,SCALAR
2118           Implements the fcntl(2) function.  You'll probably have to say
2119
2120               use Fcntl;
2121
2122           first to get the correct constant definitions.  Argument processing
2123           and value returned work just like "ioctl" below.  For example:
2124
2125               use Fcntl;
2126               my $flags = fcntl($filehandle, F_GETFL, 0)
2127                   or die "Can't fcntl F_GETFL: $!";
2128
2129           You don't have to check for "defined" on the return from "fcntl".
2130           Like "ioctl", it maps a 0 return from the system call into "0 but
2131           true" in Perl.  This string is true in boolean context and 0 in
2132           numeric context.  It is also exempt from the normal "Argument "..."
2133           isn't numeric" warnings on improper numeric conversions.
2134
2135           Note that "fcntl" raises an exception if used on a machine that
2136           doesn't implement fcntl(2).  See the Fcntl module or your fcntl(2)
2137           manpage to learn what functions are available on your system.
2138
2139           Here's an example of setting a filehandle named $REMOTE to be non-
2140           blocking at the system level.  You'll have to negotiate $| on your
2141           own, though.
2142
2143               use Fcntl qw(F_GETFL F_SETFL O_NONBLOCK);
2144
2145               my $flags = fcntl($REMOTE, F_GETFL, 0)
2146                   or die "Can't get flags for the socket: $!\n";
2147
2148               fcntl($REMOTE, F_SETFL, $flags | O_NONBLOCK)
2149                   or die "Can't set flags for the socket: $!\n";
2150
2151           Portability issues: "fcntl" in perlport.
2152
2153       __FILE__
2154           A special token that returns the name of the file in which it
2155           occurs.  It can be altered by the mechanism described at "Plain Old
2156           Comments (Not!)" in perlsyn.
2157
2158       fileno FILEHANDLE
2159       fileno DIRHANDLE
2160           Returns the file descriptor for a filehandle or directory handle,
2161           or undefined if the filehandle is not open.  If there is no real
2162           file descriptor at the OS level, as can happen with filehandles
2163           connected to memory objects via "open" with a reference for the
2164           third argument, -1 is returned.
2165
2166           This is mainly useful for constructing bitmaps for "select" and
2167           low-level POSIX tty-handling operations.  If FILEHANDLE is an
2168           expression, the value is taken as an indirect filehandle, generally
2169           its name.
2170
2171           You can use this to find out whether two handles refer to the same
2172           underlying descriptor:
2173
2174               if (fileno($this) != -1 && fileno($this) == fileno($that)) {
2175                   print "\$this and \$that are dups\n";
2176               } elsif (fileno($this) != -1 && fileno($that) != -1) {
2177                   print "\$this and \$that have different " .
2178                       "underlying file descriptors\n";
2179               } else {
2180                   print "At least one of \$this and \$that does " .
2181                       "not have a real file descriptor\n";
2182               }
2183
2184           The behavior of "fileno" on a directory handle depends on the
2185           operating system.  On a system with dirfd(3) or similar, "fileno"
2186           on a directory handle returns the underlying file descriptor
2187           associated with the handle; on systems with no such support, it
2188           returns the undefined value, and sets $! (errno).
2189
2190       flock FILEHANDLE,OPERATION
2191           Calls flock(2), or an emulation of it, on FILEHANDLE.  Returns true
2192           for success, false on failure.  Produces a fatal error if used on a
2193           machine that doesn't implement flock(2), fcntl(2) locking, or
2194           lockf(3).  "flock" is Perl's portable file-locking interface,
2195           although it locks entire files only, not records.
2196
2197           Two potentially non-obvious but traditional "flock" semantics are
2198           that it waits indefinitely until the lock is granted, and that its
2199           locks are merely advisory.  Such discretionary locks are more
2200           flexible, but offer fewer guarantees.  This means that programs
2201           that do not also use "flock" may modify files locked with "flock".
2202           See perlport, your port's specific documentation, and your system-
2203           specific local manpages for details.  It's best to assume
2204           traditional behavior if you're writing portable programs.  (But if
2205           you're not, you should as always feel perfectly free to write for
2206           your own system's idiosyncrasies (sometimes called "features").
2207           Slavish adherence to portability concerns shouldn't get in the way
2208           of your getting your job done.)
2209
2210           OPERATION is one of LOCK_SH, LOCK_EX, or LOCK_UN, possibly combined
2211           with LOCK_NB.  These constants are traditionally valued 1, 2, 8 and
2212           4, but you can use the symbolic names if you import them from the
2213           Fcntl module, either individually, or as a group using the ":flock"
2214           tag.  LOCK_SH requests a shared lock, LOCK_EX requests an exclusive
2215           lock, and LOCK_UN releases a previously requested lock.  If LOCK_NB
2216           is bitwise-or'ed with LOCK_SH or LOCK_EX, then "flock" returns
2217           immediately rather than blocking waiting for the lock; check the
2218           return status to see if you got it.
2219
2220           To avoid the possibility of miscoordination, Perl now flushes
2221           FILEHANDLE before locking or unlocking it.
2222
2223           Note that the emulation built with lockf(3) doesn't provide shared
2224           locks, and it requires that FILEHANDLE be open with write intent.
2225           These are the semantics that lockf(3) implements.  Most if not all
2226           systems implement lockf(3) in terms of fcntl(2) locking, though, so
2227           the differing semantics shouldn't bite too many people.
2228
2229           Note that the fcntl(2) emulation of flock(3) requires that
2230           FILEHANDLE be open with read intent to use LOCK_SH and requires
2231           that it be open with write intent to use LOCK_EX.
2232
2233           Note also that some versions of "flock" cannot lock things over the
2234           network; you would need to use the more system-specific "fcntl" for
2235           that.  If you like you can force Perl to ignore your system's
2236           flock(2) function, and so provide its own fcntl(2)-based emulation,
2237           by passing the switch "-Ud_flock" to the Configure program when you
2238           configure and build a new Perl.
2239
2240           Here's a mailbox appender for BSD systems.
2241
2242               # import LOCK_* and SEEK_END constants
2243               use Fcntl qw(:flock SEEK_END);
2244
2245               sub lock {
2246                   my ($fh) = @_;
2247                   flock($fh, LOCK_EX) or die "Cannot lock mailbox - $!\n";
2248                   # and, in case we're running on a very old UNIX
2249                   # variant without the modern O_APPEND semantics...
2250                   seek($fh, 0, SEEK_END) or die "Cannot seek - $!\n";
2251               }
2252
2253               sub unlock {
2254                   my ($fh) = @_;
2255                   flock($fh, LOCK_UN) or die "Cannot unlock mailbox - $!\n";
2256               }
2257
2258               open(my $mbox, ">>", "/usr/spool/mail/$ENV{'USER'}")
2259                   or die "Can't open mailbox: $!";
2260
2261               lock($mbox);
2262               print $mbox $msg,"\n\n";
2263               unlock($mbox);
2264
2265           On systems that support a real flock(2), locks are inherited across
2266           "fork" calls, whereas those that must resort to the more capricious
2267           fcntl(2) function lose their locks, making it seriously harder to
2268           write servers.
2269
2270           See also DB_File for other "flock" examples.
2271
2272           Portability issues: "flock" in perlport.
2273
2274       fork
2275           Does a fork(2) system call to create a new process running the same
2276           program at the same point.  It returns the child pid to the parent
2277           process, 0 to the child process, or "undef" if the fork is
2278           unsuccessful.  File descriptors (and sometimes locks on those
2279           descriptors) are shared, while everything else is copied.  On most
2280           systems supporting fork(2), great care has gone into making it
2281           extremely efficient (for example, using copy-on-write technology on
2282           data pages), making it the dominant paradigm for multitasking over
2283           the last few decades.
2284
2285           Perl attempts to flush all files opened for output before forking
2286           the child process, but this may not be supported on some platforms
2287           (see perlport).  To be safe, you may need to set $| ($AUTOFLUSH in
2288           English) or call the "autoflush" method of "IO::Handle" on any open
2289           handles to avoid duplicate output.
2290
2291           If you "fork" without ever waiting on your children, you will
2292           accumulate zombies.  On some systems, you can avoid this by setting
2293           $SIG{CHLD} to "IGNORE".  See also perlipc for more examples of
2294           forking and reaping moribund children.
2295
2296           Note that if your forked child inherits system file descriptors
2297           like STDIN and STDOUT that are actually connected by a pipe or
2298           socket, even if you exit, then the remote server (such as, say, a
2299           CGI script or a backgrounded job launched from a remote shell)
2300           won't think you're done.  You should reopen those to /dev/null if
2301           it's any issue.
2302
2303           On some platforms such as Windows, where the fork(2) system call is
2304           not available, Perl can be built to emulate "fork" in the Perl
2305           interpreter.  The emulation is designed, at the level of the Perl
2306           program, to be as compatible as possible with the "Unix" fork(2).
2307           However it has limitations that have to be considered in code
2308           intended to be portable.  See perlfork for more details.
2309
2310           Portability issues: "fork" in perlport.
2311
2312       format
2313           Declare a picture format for use by the "write" function.  For
2314           example:
2315
2316               format Something =
2317                   Test: @<<<<<<<< @||||| @>>>>>
2318                         $str,     $%,    '$' . int($num)
2319               .
2320
2321               $str = "widget";
2322               $num = $cost/$quantity;
2323               $~ = 'Something';
2324               write;
2325
2326           See perlform for many details and examples.
2327
2328       formline PICTURE,LIST
2329           This is an internal function used by "format"s, though you may call
2330           it, too.  It formats (see perlform) a list of values according to
2331           the contents of PICTURE, placing the output into the format output
2332           accumulator, $^A (or $ACCUMULATOR in English).  Eventually, when a
2333           "write" is done, the contents of $^A are written to some
2334           filehandle.  You could also read $^A and then set $^A back to "".
2335           Note that a format typically does one "formline" per line of form,
2336           but the "formline" function itself doesn't care how many newlines
2337           are embedded in the PICTURE.  This means that the "~" and "~~"
2338           tokens treat the entire PICTURE as a single line.  You may
2339           therefore need to use multiple formlines to implement a single
2340           record format, just like the "format" compiler.
2341
2342           Be careful if you put double quotes around the picture, because an
2343           "@" character may be taken to mean the beginning of an array name.
2344           "formline" always returns true.  See perlform for other examples.
2345
2346           If you are trying to use this instead of "write" to capture the
2347           output, you may find it easier to open a filehandle to a scalar
2348           ("open my $fh, ">", \$output") and write to that instead.
2349
2350       getc FILEHANDLE
2351       getc
2352           Returns the next character from the input file attached to
2353           FILEHANDLE, or the undefined value at end of file or if there was
2354           an error (in the latter case $! is set).  If FILEHANDLE is omitted,
2355           reads from STDIN.  This is not particularly efficient.  However, it
2356           cannot be used by itself to fetch single characters without waiting
2357           for the user to hit enter.  For that, try something more like:
2358
2359               if ($BSD_STYLE) {
2360                   system "stty cbreak </dev/tty >/dev/tty 2>&1";
2361               }
2362               else {
2363                   system "stty", '-icanon', 'eol', "\001";
2364               }
2365
2366               my $key = getc(STDIN);
2367
2368               if ($BSD_STYLE) {
2369                   system "stty -cbreak </dev/tty >/dev/tty 2>&1";
2370               }
2371               else {
2372                   system 'stty', 'icanon', 'eol', '^@'; # ASCII NUL
2373               }
2374               print "\n";
2375
2376           Determination of whether $BSD_STYLE should be set is left as an
2377           exercise to the reader.
2378
2379           The "POSIX::getattr" function can do this more portably on systems
2380           purporting POSIX compliance.  See also the "Term::ReadKey" module
2381           on CPAN.
2382
2383       getlogin
2384           This implements the C library function of the same name, which on
2385           most systems returns the current login from /etc/utmp, if any.  If
2386           it returns the empty string, use "getpwuid".
2387
2388               my $login = getlogin || getpwuid($<) || "Kilroy";
2389
2390           Do not consider "getlogin" for authentication: it is not as secure
2391           as "getpwuid".
2392
2393           Portability issues: "getlogin" in perlport.
2394
2395       getpeername SOCKET
2396           Returns the packed sockaddr address of the other end of the SOCKET
2397           connection.
2398
2399               use Socket;
2400               my $hersockaddr    = getpeername($sock);
2401               my ($port, $iaddr) = sockaddr_in($hersockaddr);
2402               my $herhostname    = gethostbyaddr($iaddr, AF_INET);
2403               my $herstraddr     = inet_ntoa($iaddr);
2404
2405       getpgrp PID
2406           Returns the current process group for the specified PID.  Use a PID
2407           of 0 to get the current process group for the current process.
2408           Will raise an exception if used on a machine that doesn't implement
2409           getpgrp(2).  If PID is omitted, returns the process group of the
2410           current process.  Note that the POSIX version of "getpgrp" does not
2411           accept a PID argument, so only "PID==0" is truly portable.
2412
2413           Portability issues: "getpgrp" in perlport.
2414
2415       getppid
2416           Returns the process id of the parent process.
2417
2418           Note for Linux users: Between v5.8.1 and v5.16.0 Perl would work
2419           around non-POSIX thread semantics the minority of Linux systems
2420           (and Debian GNU/kFreeBSD systems) that used LinuxThreads, this
2421           emulation has since been removed.  See the documentation for $$ for
2422           details.
2423
2424           Portability issues: "getppid" in perlport.
2425
2426       getpriority WHICH,WHO
2427           Returns the current priority for a process, a process group, or a
2428           user.  (See getpriority(2).)  Will raise a fatal exception if used
2429           on a machine that doesn't implement getpriority(2).
2430
2431           "WHICH" can be any of "PRIO_PROCESS", "PRIO_PGRP" or "PRIO_USER"
2432           imported from "RESOURCE CONSTANTS" in POSIX.
2433
2434           Portability issues: "getpriority" in perlport.
2435
2436       getpwnam NAME
2437       getgrnam NAME
2438       gethostbyname NAME
2439       getnetbyname NAME
2440       getprotobyname NAME
2441       getpwuid UID
2442       getgrgid GID
2443       getservbyname NAME,PROTO
2444       gethostbyaddr ADDR,ADDRTYPE
2445       getnetbyaddr ADDR,ADDRTYPE
2446       getprotobynumber NUMBER
2447       getservbyport PORT,PROTO
2448       getpwent
2449       getgrent
2450       gethostent
2451       getnetent
2452       getprotoent
2453       getservent
2454       setpwent
2455       setgrent
2456       sethostent STAYOPEN
2457       setnetent STAYOPEN
2458       setprotoent STAYOPEN
2459       setservent STAYOPEN
2460       endpwent
2461       endgrent
2462       endhostent
2463       endnetent
2464       endprotoent
2465       endservent
2466           These routines are the same as their counterparts in the system C
2467           library.  In list context, the return values from the various get
2468           routines are as follows:
2469
2470            #    0        1          2           3         4
2471            my ( $name,   $passwd,   $gid,       $members  ) = getgr*
2472            my ( $name,   $aliases,  $addrtype,  $net      ) = getnet*
2473            my ( $name,   $aliases,  $port,      $proto    ) = getserv*
2474            my ( $name,   $aliases,  $proto                ) = getproto*
2475            my ( $name,   $aliases,  $addrtype,  $length,  @addrs ) = gethost*
2476            my ( $name,   $passwd,   $uid,       $gid,     $quota,
2477               $comment,  $gcos,     $dir,       $shell,   $expire ) = getpw*
2478            #    5        6          7           8         9
2479
2480           (If the entry doesn't exist, the return value is a single
2481           meaningless true value.)
2482
2483           The exact meaning of the $gcos field varies but it usually contains
2484           the real name of the user (as opposed to the login name) and other
2485           information pertaining to the user.  Beware, however, that in many
2486           system users are able to change this information and therefore it
2487           cannot be trusted and therefore the $gcos is tainted (see perlsec).
2488           The $passwd and $shell, user's encrypted password and login shell,
2489           are also tainted, for the same reason.
2490
2491           In scalar context, you get the name, unless the function was a
2492           lookup by name, in which case you get the other thing, whatever it
2493           is.  (If the entry doesn't exist you get the undefined value.)  For
2494           example:
2495
2496               my $uid   = getpwnam($name);
2497               my $name  = getpwuid($num);
2498               my $name  = getpwent();
2499               my $gid   = getgrnam($name);
2500               my $name  = getgrgid($num);
2501               my $name  = getgrent();
2502               # etc.
2503
2504           In getpw*() the fields $quota, $comment, and $expire are special in
2505           that they are unsupported on many systems.  If the $quota is
2506           unsupported, it is an empty scalar.  If it is supported, it usually
2507           encodes the disk quota.  If the $comment field is unsupported, it
2508           is an empty scalar.  If it is supported it usually encodes some
2509           administrative comment about the user.  In some systems the $quota
2510           field may be $change or $age, fields that have to do with password
2511           aging.  In some systems the $comment field may be $class.  The
2512           $expire field, if present, encodes the expiration period of the
2513           account or the password.  For the availability and the exact
2514           meaning of these fields in your system, please consult getpwnam(3)
2515           and your system's pwd.h file.  You can also find out from within
2516           Perl what your $quota and $comment fields mean and whether you have
2517           the $expire field by using the "Config" module and the values
2518           "d_pwquota", "d_pwage", "d_pwchange", "d_pwcomment", and
2519           "d_pwexpire".  Shadow password files are supported only if your
2520           vendor has implemented them in the intuitive fashion that calling
2521           the regular C library routines gets the shadow versions if you're
2522           running under privilege or if there exists the shadow(3) functions
2523           as found in System V (this includes Solaris and Linux).  Those
2524           systems that implement a proprietary shadow password facility are
2525           unlikely to be supported.
2526
2527           The $members value returned by getgr*() is a space-separated list
2528           of the login names of the members of the group.
2529
2530           For the gethost*() functions, if the "h_errno" variable is
2531           supported in C, it will be returned to you via $? if the function
2532           call fails.  The @addrs value returned by a successful call is a
2533           list of raw addresses returned by the corresponding library call.
2534           In the Internet domain, each address is four bytes long; you can
2535           unpack it by saying something like:
2536
2537               my ($w,$x,$y,$z) = unpack('W4',$addr[0]);
2538
2539           The Socket library makes this slightly easier:
2540
2541               use Socket;
2542               my $iaddr = inet_aton("127.1"); # or whatever address
2543               my $name  = gethostbyaddr($iaddr, AF_INET);
2544
2545               # or going the other way
2546               my $straddr = inet_ntoa($iaddr);
2547
2548           In the opposite way, to resolve a hostname to the IP address you
2549           can write this:
2550
2551               use Socket;
2552               my $packed_ip = gethostbyname("www.perl.org");
2553               my $ip_address;
2554               if (defined $packed_ip) {
2555                   $ip_address = inet_ntoa($packed_ip);
2556               }
2557
2558           Make sure "gethostbyname" is called in SCALAR context and that its
2559           return value is checked for definedness.
2560
2561           The "getprotobynumber" function, even though it only takes one
2562           argument, has the precedence of a list operator, so beware:
2563
2564               getprotobynumber $number eq 'icmp'   # WRONG
2565               getprotobynumber($number eq 'icmp')  # actually means this
2566               getprotobynumber($number) eq 'icmp'  # better this way
2567
2568           If you get tired of remembering which element of the return list
2569           contains which return value, by-name interfaces are provided in
2570           standard modules: "File::stat", "Net::hostent", "Net::netent",
2571           "Net::protoent", "Net::servent", "Time::gmtime", "Time::localtime",
2572           and "User::grent".  These override the normal built-ins, supplying
2573           versions that return objects with the appropriate names for each
2574           field.  For example:
2575
2576              use File::stat;
2577              use User::pwent;
2578              my $is_his = (stat($filename)->uid == pwent($whoever)->uid);
2579
2580           Even though it looks as though they're the same method calls (uid),
2581           they aren't, because a "File::stat" object is different from a
2582           "User::pwent" object.
2583
2584           Many of these functions are not safe in a multi-threaded
2585           environment where more than one thread can be using them.  In
2586           particular, functions like "getpwent()" iterate per-process and not
2587           per-thread, so if two threads are simultaneously iterating, neither
2588           will get all the records.
2589
2590           Some systems have thread-safe versions of some of the functions,
2591           such as "getpwnam_r()" instead of "getpwnam()".  There, Perl
2592           automatically and invisibly substitutes the thread-safe version,
2593           without notice.  This means that code that safely runs on some
2594           systems can fail on others that lack the thread-safe versions.
2595
2596           Portability issues: "getpwnam" in perlport to "endservent" in
2597           perlport.
2598
2599       getsockname SOCKET
2600           Returns the packed sockaddr address of this end of the SOCKET
2601           connection, in case you don't know the address because you have
2602           several different IPs that the connection might have come in on.
2603
2604               use Socket;
2605               my $mysockaddr = getsockname($sock);
2606               my ($port, $myaddr) = sockaddr_in($mysockaddr);
2607               printf "Connect to %s [%s]\n",
2608                  scalar gethostbyaddr($myaddr, AF_INET),
2609                  inet_ntoa($myaddr);
2610
2611       getsockopt SOCKET,LEVEL,OPTNAME
2612           Queries the option named OPTNAME associated with SOCKET at a given
2613           LEVEL.  Options may exist at multiple protocol levels depending on
2614           the socket type, but at least the uppermost socket level SOL_SOCKET
2615           (defined in the "Socket" module) will exist.  To query options at
2616           another level the protocol number of the appropriate protocol
2617           controlling the option should be supplied.  For example, to
2618           indicate that an option is to be interpreted by the TCP protocol,
2619           LEVEL should be set to the protocol number of TCP, which you can
2620           get using "getprotobyname".
2621
2622           The function returns a packed string representing the requested
2623           socket option, or "undef" on error, with the reason for the error
2624           placed in $!.  Just what is in the packed string depends on LEVEL
2625           and OPTNAME; consult getsockopt(2) for details.  A common case is
2626           that the option is an integer, in which case the result is a packed
2627           integer, which you can decode using "unpack" with the "i" (or "I")
2628           format.
2629
2630           Here's an example to test whether Nagle's algorithm is enabled on a
2631           socket:
2632
2633               use Socket qw(:all);
2634
2635               defined(my $tcp = getprotobyname("tcp"))
2636                   or die "Could not determine the protocol number for tcp";
2637               # my $tcp = IPPROTO_TCP; # Alternative
2638               my $packed = getsockopt($socket, $tcp, TCP_NODELAY)
2639                   or die "getsockopt TCP_NODELAY: $!";
2640               my $nodelay = unpack("I", $packed);
2641               print "Nagle's algorithm is turned ",
2642                      $nodelay ? "off\n" : "on\n";
2643
2644           Portability issues: "getsockopt" in perlport.
2645
2646       glob EXPR
2647       glob
2648           In list context, returns a (possibly empty) list of filename
2649           expansions on the value of EXPR such as the Unix shell Bash would
2650           do. In scalar context, glob iterates through such filename
2651           expansions, returning "undef" when the list is exhausted. If EXPR
2652           is omitted, $_ is used.
2653
2654               # List context
2655               my @txt_files  = glob("*.txt");
2656               my @perl_files = glob("*.pl *.pm");
2657
2658               # Scalar context
2659               while (my $file = glob("*.mp3")) {
2660                   # Do stuff
2661               }
2662
2663           Glob also supports an alternate syntax using "<" ">" as delimiters.
2664           While this syntax is supported, it is recommended that you use
2665           "glob" instead as it is more readable and searchable.
2666
2667               my @txt_files  = <"*.txt">;
2668
2669           If you need case insensitive file globbing that can be achieved
2670           using the ":nocase" parameter of the "bsd_glob" module.
2671
2672               use File::Glob qw(:globally :nocase);
2673
2674                   my @txt = glob("readme*"); # README readme.txt Readme.md
2675
2676           Note that "glob" splits its arguments on whitespace and treats each
2677           segment as separate pattern.  As such, "glob("*.c *.h")" matches
2678           all files with a .c or .h extension.  The expression "glob(".* *")"
2679           matches all files in the current working directory.  If you want to
2680           glob filenames that might contain whitespace, you'll have to use
2681           extra quotes around the spacey filename to protect it.  For
2682           example, to glob filenames that have an "e" followed by a space
2683           followed by an "f", use one of:
2684
2685               my @spacies = <"*e f*">;
2686               my @spacies = glob('"*e f*"');
2687               my @spacies = glob(q("*e f*"));
2688
2689           If you had to get a variable through, you could do this:
2690
2691               my @spacies = glob("'*${var}e f*'");
2692               my @spacies = glob(qq("*${var}e f*"));
2693
2694           If non-empty braces are the only wildcard characters used in the
2695           "glob", no filenames are matched, but potentially many strings are
2696           returned.  For example, this produces nine strings, one for each
2697           pairing of fruits and colors:
2698
2699               my @many = glob("{apple,tomato,cherry}={green,yellow,red}");
2700
2701           This operator is implemented using the standard "File::Glob"
2702           extension.  See "bsd_glob" for details, including "bsd_glob", which
2703           does not treat whitespace as a pattern separator.
2704
2705           If a "glob" expression is used as the condition of a "while" or
2706           "for" loop, then it will be implicitly assigned to $_.  If either a
2707           "glob" expression or an explicit assignment of a "glob" expression
2708           to a scalar is used as a "while"/"for" condition, then the
2709           condition actually tests for definedness of the expression's value,
2710           not for its regular truth value.
2711
2712           Internal implemenation details:
2713
2714           This is the internal function implementing the "<*.c>" operator,
2715           but you can use it directly. The "<*.c>" operator is discussed in
2716           more detail in "I/O Operators" in perlop.
2717
2718           Portability issues: "glob" in perlport.
2719
2720       gmtime EXPR
2721       gmtime
2722           Works just like "localtime", but the returned values are localized
2723           for the standard Greenwich time zone.
2724
2725           Note: When called in list context, $isdst, the last value returned
2726           by gmtime, is always 0.  There is no Daylight Saving Time in GMT.
2727
2728           Portability issues: "gmtime" in perlport.
2729
2730       goto LABEL
2731       goto EXPR
2732       goto &NAME
2733           The "goto LABEL" form finds the statement labeled with LABEL and
2734           resumes execution there.  It can't be used to get out of a block or
2735           subroutine given to "sort".  It can be used to go almost anywhere
2736           else within the dynamic scope, including out of subroutines, but
2737           it's usually better to use some other construct such as "last" or
2738           "die".  The author of Perl has never felt the need to use this form
2739           of "goto" (in Perl, that is; C is another matter).  (The difference
2740           is that C does not offer named loops combined with loop control.
2741           Perl does, and this replaces most structured uses of "goto" in
2742           other languages.)
2743
2744           The "goto EXPR" form expects to evaluate "EXPR" to a code reference
2745           or a label name.  If it evaluates to a code reference, it will be
2746           handled like "goto &NAME", below.  This is especially useful for
2747           implementing tail recursion via "goto __SUB__".
2748
2749           If the expression evaluates to a label name, its scope will be
2750           resolved dynamically.  This allows for computed "goto"s per
2751           FORTRAN, but isn't necessarily recommended if you're optimizing for
2752           maintainability:
2753
2754               goto ("FOO", "BAR", "GLARCH")[$i];
2755
2756           As shown in this example, "goto EXPR" is exempt from the "looks
2757           like a function" rule.  A pair of parentheses following it does not
2758           (necessarily) delimit its argument.  "goto("NE")."XT"" is
2759           equivalent to "goto NEXT".  Also, unlike most named operators, this
2760           has the same precedence as assignment.
2761
2762           Use of "goto LABEL" or "goto EXPR" to jump into a construct is
2763           deprecated and will issue a warning.  Even then, it may not be used
2764           to go into any construct that requires initialization, such as a
2765           subroutine, a "foreach" loop, or a "given" block.  In general, it
2766           may not be used to jump into the parameter of a binary or list
2767           operator, but it may be used to jump into the first parameter of a
2768           binary operator.  (The "=" assignment operator's "first" operand is
2769           its right-hand operand.)  It also can't be used to go into a
2770           construct that is optimized away.
2771
2772           The "goto &NAME" form is quite different from the other forms of
2773           "goto".  In fact, it isn't a goto in the normal sense at all, and
2774           doesn't have the stigma associated with other gotos.  Instead, it
2775           exits the current subroutine (losing any changes set by "local")
2776           and immediately calls in its place the named subroutine using the
2777           current value of @_.  This is used by "AUTOLOAD" subroutines that
2778           wish to load another subroutine and then pretend that the other
2779           subroutine had been called in the first place (except that any
2780           modifications to @_ in the current subroutine are propagated to the
2781           other subroutine.) After the "goto", not even "caller" will be able
2782           to tell that this routine was called first.
2783
2784           NAME needn't be the name of a subroutine; it can be a scalar
2785           variable containing a code reference or a block that evaluates to a
2786           code reference.
2787
2788       grep BLOCK LIST
2789       grep EXPR,LIST
2790           This is similar in spirit to, but not the same as, grep(1) and its
2791           relatives.  In particular, it is not limited to using regular
2792           expressions.
2793
2794           Evaluates the BLOCK or EXPR for each element of LIST (locally
2795           setting $_ to each element) and returns the list value consisting
2796           of those elements for which the expression evaluated to true.  In
2797           scalar context, returns the number of times the expression was
2798           true.
2799
2800               my @foo = grep(!/^#/, @bar);    # weed out comments
2801
2802           or equivalently,
2803
2804               my @foo = grep {!/^#/} @bar;    # weed out comments
2805
2806           Note that $_ is an alias to the list value, so it can be used to
2807           modify the elements of the LIST.  While this is useful and
2808           supported, it can cause bizarre results if the elements of LIST are
2809           not variables.  Similarly, grep returns aliases into the original
2810           list, much as a for loop's index variable aliases the list
2811           elements.  That is, modifying an element of a list returned by grep
2812           (for example, in a "foreach", "map" or another "grep") actually
2813           modifies the element in the original list.  This is usually
2814           something to be avoided when writing clear code.
2815
2816           See also "map" for a list composed of the results of the BLOCK or
2817           EXPR.
2818
2819       hex EXPR
2820       hex Interprets EXPR as a hex string and returns the corresponding
2821           numeric value.  If EXPR is omitted, uses $_.
2822
2823               print hex '0xAf'; # prints '175'
2824               print hex 'aF';   # same
2825               $valid_input =~ /\A(?:0?[xX])?(?:_?[0-9a-fA-F])*\z/
2826
2827           A hex string consists of hex digits and an optional "0x" or "x"
2828           prefix.  Each hex digit may be preceded by a single underscore,
2829           which will be ignored.  Any other character triggers a warning and
2830           causes the rest of the string to be ignored (even leading
2831           whitespace, unlike "oct").  Only integers can be represented, and
2832           integer overflow triggers a warning.
2833
2834           To convert strings that might start with any of 0, "0x", or "0b",
2835           see "oct".  To present something as hex, look into "printf",
2836           "sprintf", and "unpack".
2837
2838       import LIST
2839           There is no builtin "import" function.  It is just an ordinary
2840           method (subroutine) defined (or inherited) by modules that wish to
2841           export names to another module.  The "use" function calls the
2842           "import" method for the package used.  See also "use", perlmod, and
2843           Exporter.
2844
2845       index STR,SUBSTR,POSITION
2846       index STR,SUBSTR
2847           The index function searches for one string within another, but
2848           without the wildcard-like behavior of a full regular-expression
2849           pattern match.  It returns the position of the first occurrence of
2850           SUBSTR in STR at or after POSITION.  If POSITION is omitted, starts
2851           searching from the beginning of the string.  POSITION before the
2852           beginning of the string or after its end is treated as if it were
2853           the beginning or the end, respectively.  POSITION and the return
2854           value are based at zero.  If the substring is not found, "index"
2855           returns -1.
2856
2857           Find characters or strings:
2858
2859               index("Perl is great", "P");     # Returns 0
2860               index("Perl is great", "g");     # Returns 8
2861               index("Perl is great", "great"); # Also returns 8
2862
2863           Attempting to find something not there:
2864
2865               index("Perl is great", "Z");     # Returns -1 (not found)
2866
2867           Using an offset to find the second occurrence:
2868
2869               index("Perl is great", "e", 5);  # Returns 10
2870
2871       int EXPR
2872       int Returns the integer portion of EXPR.  If EXPR is omitted, uses $_.
2873           You should not use this function for rounding: one because it
2874           truncates towards 0, and two because machine representations of
2875           floating-point numbers can sometimes produce counterintuitive
2876           results.  For example, "int(-6.725/0.025)" produces -268 rather
2877           than the correct -269; that's because it's really more like
2878           -268.99999999999994315658 instead.  Usually, the "sprintf",
2879           "printf", or the "POSIX::floor" and "POSIX::ceil" functions will
2880           serve you better than will "int".
2881
2882       ioctl FILEHANDLE,FUNCTION,SCALAR
2883           Implements the ioctl(2) function.  You'll probably first have to
2884           say
2885
2886               require "sys/ioctl.ph";  # probably in
2887                                        # $Config{archlib}/sys/ioctl.ph
2888
2889           to get the correct function definitions.  If sys/ioctl.ph doesn't
2890           exist or doesn't have the correct definitions you'll have to roll
2891           your own, based on your C header files such as <sys/ioctl.h>.
2892           (There is a Perl script called h2ph that comes with the Perl kit
2893           that may help you in this, but it's nontrivial.)  SCALAR will be
2894           read and/or written depending on the FUNCTION; a C pointer to the
2895           string value of SCALAR will be passed as the third argument of the
2896           actual "ioctl" call.  (If SCALAR has no string value but does have
2897           a numeric value, that value will be passed rather than a pointer to
2898           the string value.  To guarantee this to be true, add a 0 to the
2899           scalar before using it.)  The "pack" and "unpack" functions may be
2900           needed to manipulate the values of structures used by "ioctl".
2901
2902           The return value of "ioctl" (and "fcntl") is as follows:
2903
2904               if OS returns:      then Perl returns:
2905                   -1               undefined value
2906                    0              string "0 but true"
2907               anything else           that number
2908
2909           Thus Perl returns true on success and false on failure, yet you can
2910           still easily determine the actual value returned by the operating
2911           system:
2912
2913               my $retval = ioctl(...) || -1;
2914               printf "System returned %d\n", $retval;
2915
2916           The special string "0 but true" is exempt from "Argument "..."
2917           isn't numeric" warnings on improper numeric conversions.
2918
2919           Portability issues: "ioctl" in perlport.
2920
2921       join EXPR,LIST
2922           Joins the separate strings of LIST into a single string with fields
2923           separated by the value of EXPR, and returns that new string.
2924           Example:
2925
2926              my $rec = join(':', $login,$passwd,$uid,$gid,$gcos,$home,$shell);
2927
2928           Beware that unlike "split", "join" doesn't take a pattern as its
2929           first argument.  Compare "split".
2930
2931       keys HASH
2932       keys ARRAY
2933           Called in list context, returns a list consisting of all the keys
2934           of the named hash, or in Perl 5.12 or later only, the indices of an
2935           array.  Perl releases prior to 5.12 will produce a syntax error if
2936           you try to use an array argument.  In scalar context, returns the
2937           number of keys or indices.
2938
2939           Hash entries are returned in an apparently random order.  The
2940           actual random order is specific to a given hash; the exact same
2941           series of operations on two hashes may result in a different order
2942           for each hash.  Any insertion into the hash may change the order,
2943           as will any deletion, with the exception that the most recent key
2944           returned by "each" or "keys" may be deleted without changing the
2945           order.  So long as a given hash is unmodified you may rely on
2946           "keys", "values" and "each" to repeatedly return the same order as
2947           each other.  See "Algorithmic Complexity Attacks" in perlsec for
2948           details on why hash order is randomized.  Aside from the guarantees
2949           provided here the exact details of Perl's hash algorithm and the
2950           hash traversal order are subject to change in any release of Perl.
2951           Tied hashes may behave differently to Perl's hashes with respect to
2952           changes in order on insertion and deletion of items.
2953
2954           As a side effect, calling "keys" resets the internal iterator of
2955           the HASH or ARRAY (see "each") before yielding the keys.  In
2956           particular, calling "keys" in void context resets the iterator with
2957           no other overhead.
2958
2959           Here is yet another way to print your environment:
2960
2961               my @keys = keys %ENV;
2962               my @values = values %ENV;
2963               while (@keys) {
2964                   print pop(@keys), '=', pop(@values), "\n";
2965               }
2966
2967           or how about sorted by key:
2968
2969               foreach my $key (sort(keys %ENV)) {
2970                   print $key, '=', $ENV{$key}, "\n";
2971               }
2972
2973           The returned values are copies of the original keys in the hash, so
2974           modifying them will not affect the original hash.  Compare
2975           "values".
2976
2977           To sort a hash by value, you'll need to use a "sort" function.
2978           Here's a descending numeric sort of a hash by its values:
2979
2980               foreach my $key (sort { $hash{$b} <=> $hash{$a} } keys %hash) {
2981                   printf "%4d %s\n", $hash{$key}, $key;
2982               }
2983
2984           Used as an lvalue, "keys" allows you to increase the number of hash
2985           buckets allocated for the given hash.  This can gain you a measure
2986           of efficiency if you know the hash is going to get big.  (This is
2987           similar to pre-extending an array by assigning a larger number to
2988           $#array.)  If you say
2989
2990               keys %hash = 200;
2991
2992           then %hash will have at least 200 buckets allocated for it--256 of
2993           them, in fact, since it rounds up to the next power of two.  These
2994           buckets will be retained even if you do "%hash = ()", use "undef
2995           %hash" if you want to free the storage while %hash is still in
2996           scope.  You can't shrink the number of buckets allocated for the
2997           hash using "keys" in this way (but you needn't worry about doing
2998           this by accident, as trying has no effect).  "keys @array" in an
2999           lvalue context is a syntax error.
3000
3001           Starting with Perl 5.14, an experimental feature allowed "keys" to
3002           take a scalar expression. This experiment has been deemed
3003           unsuccessful, and was removed as of Perl 5.24.
3004
3005           To avoid confusing would-be users of your code who are running
3006           earlier versions of Perl with mysterious syntax errors, put this
3007           sort of thing at the top of your file to signal that your code will
3008           work only on Perls of a recent vintage:
3009
3010               use v5.12;  # so keys/values/each work on arrays
3011
3012           See also "each", "values", and "sort".
3013
3014       kill SIGNAL, LIST
3015       kill SIGNAL
3016           Sends a signal to a list of processes.  Returns the number of
3017           arguments that were successfully used to signal (which is not
3018           necessarily the same as the number of processes actually killed,
3019           e.g. where a process group is killed).
3020
3021               my $cnt = kill 'HUP', $child1, $child2;
3022               kill 'KILL', @goners;
3023
3024           SIGNAL may be either a signal name (a string) or a signal number.
3025           A signal name may start with a "SIG" prefix, thus "FOO" and
3026           "SIGFOO" refer to the same signal.  The string form of SIGNAL is
3027           recommended for portability because the same signal may have
3028           different numbers in different operating systems.
3029
3030           A list of signal names supported by the current platform can be
3031           found in $Config{sig_name}, which is provided by the "Config"
3032           module.  See Config for more details.
3033
3034           A negative signal name is the same as a negative signal number,
3035           killing process groups instead of processes.  For example, "kill
3036           '-KILL', $pgrp" and "kill -9, $pgrp" will send "SIGKILL" to the
3037           entire process group specified.  That means you usually want to use
3038           positive not negative signals.
3039
3040           If SIGNAL is either the number 0 or the string "ZERO" (or
3041           "SIGZERO"), no signal is sent to the process, but "kill" checks
3042           whether it's possible to send a signal to it (that means, to be
3043           brief, that the process is owned by the same user, or we are the
3044           super-user).  This is useful to check that a child process is still
3045           alive (even if only as a zombie) and hasn't changed its UID.  See
3046           perlport for notes on the portability of this construct.
3047
3048           The behavior of kill when a PROCESS number is zero or negative
3049           depends on the operating system.  For example, on POSIX-conforming
3050           systems, zero will signal the current process group, -1 will signal
3051           all processes, and any other negative PROCESS number will act as a
3052           negative signal number and kill the entire process group specified.
3053
3054           If both the SIGNAL and the PROCESS are negative, the results are
3055           undefined.  A warning may be produced in a future version.
3056
3057           See "Signals" in perlipc for more details.
3058
3059           On some platforms such as Windows where the fork(2) system call is
3060           not available, Perl can be built to emulate "fork" at the
3061           interpreter level.  This emulation has limitations related to kill
3062           that have to be considered, for code running on Windows and in code
3063           intended to be portable.
3064
3065           See perlfork for more details.
3066
3067           If there is no LIST of processes, no signal is sent, and the return
3068           value is 0.  This form is sometimes used, however, because it
3069           causes tainting checks to be run, if your perl support taint
3070           checks.  But see "Laundering and Detecting Tainted Data" in
3071           perlsec.
3072
3073           Portability issues: "kill" in perlport.
3074
3075       last LABEL
3076       last EXPR
3077       last
3078           The "last" command is like the "break" statement in C (as used in
3079           loops); it immediately exits the loop in question.  If the LABEL is
3080           omitted, the command refers to the innermost enclosing loop.  The
3081           "last EXPR" form, available starting in Perl 5.18.0, allows a label
3082           name to be computed at run time, and is otherwise identical to
3083           "last LABEL".  The "continue" block, if any, is not executed:
3084
3085               LINE: while (<STDIN>) {
3086                   last LINE if /^$/;  # exit when done with header
3087                   #...
3088               }
3089
3090           "last" cannot return a value from a block that typically returns a
3091           value, such as "eval {}", "sub {}", or "do {}". It will perform its
3092           flow control behavior, which precludes any return value. It should
3093           not be used to exit a "grep" or "map" operation.
3094
3095           Note that a block by itself is semantically identical to a loop
3096           that executes once.  Thus "last" can be used to effect an early
3097           exit out of such a block.
3098
3099           See also "continue" for an illustration of how "last", "next", and
3100           "redo" work.
3101
3102           Unlike most named operators, this has the same precedence as
3103           assignment.  It is also exempt from the looks-like-a-function rule,
3104           so "last ("foo")."bar"" will cause "bar" to be part of the argument
3105           to "last".
3106
3107       lc EXPR
3108       lc  Returns a lowercased version of EXPR.  If EXPR is omitted, uses $_.
3109
3110               my $str = lc("Perl is GREAT"); # "perl is great"
3111
3112           What gets returned depends on several factors:
3113
3114           If "use bytes" is in effect:
3115               The results follow ASCII rules.  Only the characters "A-Z"
3116               change, to "a-z" respectively.
3117
3118           Otherwise, if "use locale" for "LC_CTYPE" is in effect:
3119               Respects current "LC_CTYPE" locale for code points < 256; and
3120               uses Unicode rules for the remaining code points (this last can
3121               only happen if the UTF8 flag is also set).  See perllocale.
3122
3123               Starting in v5.20, Perl uses full Unicode rules if the locale
3124               is UTF-8.  Otherwise, there is a deficiency in this scheme,
3125               which is that case changes that cross the 255/256 boundary are
3126               not well-defined.  For example, the lower case of LATIN CAPITAL
3127               LETTER SHARP S (U+1E9E) in Unicode rules is U+00DF (on ASCII
3128               platforms).   But under "use locale" (prior to v5.20 or not a
3129               UTF-8 locale), the lower case of U+1E9E is itself, because 0xDF
3130               may not be LATIN SMALL LETTER SHARP S in the current locale,
3131               and Perl has no way of knowing if that character even exists in
3132               the locale, much less what code point it is.  Perl returns a
3133               result that is above 255 (almost always the input character
3134               unchanged), for all instances (and there aren't many) where the
3135               255/256 boundary would otherwise be crossed; and starting in
3136               v5.22, it raises a locale warning.
3137
3138           Otherwise, If EXPR has the UTF8 flag set:
3139               Unicode rules are used for the case change.
3140
3141           Otherwise, if "use feature 'unicode_strings'" or "use locale
3142           ':not_characters'" is in effect:
3143               Unicode rules are used for the case change.
3144
3145           Otherwise:
3146               ASCII rules are used for the case change.  The lowercase of any
3147               character outside the ASCII range is the character itself.
3148
3149           Note: This is the internal function implementing the "\L" escape in
3150           double-quoted strings.
3151
3152               my $str = "Perl is \LGREAT\E"; # "Perl is great"
3153
3154       lcfirst EXPR
3155       lcfirst
3156           Returns the value of EXPR with the first character lowercased.
3157           This is the internal function implementing the "\l" escape in
3158           double-quoted strings.
3159
3160           If EXPR is omitted, uses $_.
3161
3162           This function behaves the same way under various pragmas, such as
3163           in a locale, as "lc" does.
3164
3165       length EXPR
3166       length
3167           Returns the length in characters of the value of EXPR.  If EXPR is
3168           omitted, returns the length of $_.  If EXPR is undefined, returns
3169           "undef".
3170
3171           This function cannot be used on an entire array or hash to find out
3172           how many elements these have.  For that, use "scalar @array" and
3173           "scalar keys %hash", respectively.
3174
3175           Like all Perl character operations, "length" normally deals in
3176           logical characters, not physical bytes.  For how many bytes a
3177           string encoded as UTF-8 would take up, use
3178           "length(Encode::encode('UTF-8', EXPR))" (you'll have to "use
3179           Encode" first).  See Encode and perlunicode.
3180
3181       __LINE__
3182           A special token that compiles to the current line number.  It can
3183           be altered by the mechanism described at "Plain Old Comments
3184           (Not!)" in perlsyn.
3185
3186       link OLDFILE,NEWFILE
3187           Creates a new filename linked to the old filename.  Returns true
3188           for success, false otherwise.
3189
3190           Portability issues: "link" in perlport.
3191
3192       listen SOCKET,QUEUESIZE
3193           Does the same thing that the listen(2) system call does.  Returns
3194           true if it succeeded, false otherwise.  See the example in
3195           "Sockets: Client/Server Communication" in perlipc.
3196
3197       local EXPR
3198           You really probably want to be using "my" instead, because "local"
3199           isn't what most people think of as "local".  See "Private Variables
3200           via my()" in perlsub for details.
3201
3202           A local modifies the listed variables to be local to the enclosing
3203           block, file, or eval.  If more than one value is listed, the list
3204           must be placed in parentheses.  See "Temporary Values via local()"
3205           in perlsub for details, including issues with tied arrays and
3206           hashes.
3207
3208           The "delete local EXPR" construct can also be used to localize the
3209           deletion of array/hash elements to the current block.  See
3210           "Localized deletion of elements of composite types" in perlsub.
3211
3212       localtime EXPR
3213       localtime
3214           Converts a time as returned by the time function to a 9-element
3215           list with the time analyzed for the local time zone.  Typically
3216           used as follows:
3217
3218               #     0    1    2     3     4    5     6     7     8
3219               my ($sec,$min,$hour,$mday,$mon,$year,$wday,$yday,$isdst) =
3220                                                           localtime(time);
3221
3222           All list elements are numeric and come straight out of the C
3223           `struct tm'.  $sec, $min, and $hour are the seconds, minutes, and
3224           hours of the specified time.
3225
3226           $mday is the day of the month and $mon the month in the range
3227           0..11, with 0 indicating January and 11 indicating December.  This
3228           makes it easy to get a month name from a list:
3229
3230               my @abbr = qw(Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec);
3231               print "$abbr[$mon] $mday";
3232               # $mon=9, $mday=18 gives "Oct 18"
3233
3234           $year contains the number of years since 1900.  To get a 4-digit
3235           year write:
3236
3237               $year += 1900;
3238
3239           To get the last two digits of the year (e.g., "01" in 2001) do:
3240
3241               $year = sprintf("%02d", $year % 100);
3242
3243           $wday is the day of the week, with 0 indicating Sunday and 3
3244           indicating Wednesday.  $yday is the day of the year, in the range
3245           0..364 (or 0..365 in leap years.)
3246
3247           $isdst is true if the specified time occurs when Daylight Saving
3248           Time is in effect, false otherwise.
3249
3250           If EXPR is omitted, "localtime" uses the current time (as returned
3251           by "time").
3252
3253           In scalar context, "localtime" returns the ctime(3) value:
3254
3255            my $now_string = localtime;  # e.g., "Thu Oct 13 04:54:34 1994"
3256
3257           This scalar value is always in English, and is not locale-
3258           dependent.  To get similar but locale-dependent date strings, try
3259           for example:
3260
3261            use POSIX qw(strftime);
3262            my $now_string = strftime "%a %b %e %H:%M:%S %Y", localtime;
3263            # or for GMT formatted appropriately for your locale:
3264            my $now_string = strftime "%a %b %e %H:%M:%S %Y", gmtime;
3265
3266           C$now_string> will be formatted according to the current LC_TIME
3267           locale the program or thread is running in.  See perllocale for how
3268           to set up and change that locale.  Note that %a and %b, the short
3269           forms of the day of the week and the month of the year, may not
3270           necessarily be three characters wide.
3271
3272           The Time::gmtime and Time::localtime modules provide a convenient,
3273           by-name access mechanism to the "gmtime" and "localtime" functions,
3274           respectively.
3275
3276           For a comprehensive date and time representation look at the
3277           DateTime module on CPAN.
3278
3279           For GMT instead of local time use the "gmtime" builtin.
3280
3281           See also the "Time::Local" module (for converting seconds, minutes,
3282           hours, and such back to the integer value returned by "time"), and
3283           the POSIX module's "mktime" function.
3284
3285           Portability issues: "localtime" in perlport.
3286
3287       lock THING
3288           This function places an advisory lock on a shared variable or
3289           referenced object contained in THING until the lock goes out of
3290           scope.
3291
3292           The value returned is the scalar itself, if the argument is a
3293           scalar, or a reference, if the argument is a hash, array or
3294           subroutine.
3295
3296           "lock" is a "weak keyword"; this means that if you've defined a
3297           function by this name (before any calls to it), that function will
3298           be called instead.  If you are not under "use threads::shared" this
3299           does nothing.  See threads::shared.
3300
3301       log EXPR
3302       log Returns the natural logarithm (base e) of EXPR.  If EXPR is
3303           omitted, returns the log of $_.  To get the log of another base,
3304           use basic algebra: The base-N log of a number is equal to the
3305           natural log of that number divided by the natural log of N.  For
3306           example:
3307
3308               sub log10 {
3309                   my $n = shift;
3310                   return log($n)/log(10);
3311               }
3312
3313           See also "exp" for the inverse operation.
3314
3315       lstat FILEHANDLE
3316       lstat EXPR
3317       lstat DIRHANDLE
3318       lstat
3319           Does the same thing as the "stat" function (including setting the
3320           special "_" filehandle) but stats a symbolic link instead of the
3321           file the symbolic link points to.  If symbolic links are
3322           unimplemented on your system, a normal "stat" is done.  For much
3323           more detailed information, please see the documentation for "stat".
3324
3325           If EXPR is omitted, stats $_.
3326
3327           Portability issues: "lstat" in perlport.
3328
3329       m// The match operator.  See "Regexp Quote-Like Operators" in perlop.
3330
3331       map BLOCK LIST
3332       map EXPR,LIST
3333           Evaluates the BLOCK or EXPR for each element of LIST (locally
3334           setting $_ to each element) and composes a list of the results of
3335           each such evaluation.  Each element of LIST may produce zero, one,
3336           or more elements in the generated list, so the number of elements
3337           in the generated list may differ from that in LIST.  In scalar
3338           context, returns the total number of elements so generated.  In
3339           list context, returns the generated list.
3340
3341               my @chars = map(chr, @numbers);
3342
3343           translates a list of numbers to the corresponding characters.
3344
3345               my @squares = map { $_ * $_ } @numbers;
3346
3347           translates a list of numbers to their squared values.
3348
3349               my @squares = map { $_ > 5 ? ($_ * $_) : () } @numbers;
3350
3351           shows that number of returned elements can differ from the number
3352           of input elements.  To omit an element, return an empty list ().
3353           This could also be achieved by writing
3354
3355               my @squares = map { $_ * $_ } grep { $_ > 5 } @numbers;
3356
3357           which makes the intention more clear.
3358
3359           Map always returns a list, which can be assigned to a hash such
3360           that the elements become key/value pairs.  See perldata for more
3361           details.
3362
3363               my %hash = map { get_a_key_for($_) => $_ } @array;
3364
3365           is just a funny way to write
3366
3367               my %hash;
3368               foreach (@array) {
3369                   $hash{get_a_key_for($_)} = $_;
3370               }
3371
3372           Note that $_ is an alias to the list value, so it can be used to
3373           modify the elements of the LIST.  While this is useful and
3374           supported, it can cause bizarre results if the elements of LIST are
3375           not variables.  Using a regular "foreach" loop for this purpose
3376           would be clearer in most cases.  See also "grep" for a list
3377           composed of those items of the original list for which the BLOCK or
3378           EXPR evaluates to true.
3379
3380           "{" starts both hash references and blocks, so "map { ..." could be
3381           either the start of map BLOCK LIST or map EXPR, LIST.  Because Perl
3382           doesn't look ahead for the closing "}" it has to take a guess at
3383           which it's dealing with based on what it finds just after the "{".
3384           Usually it gets it right, but if it doesn't it won't realize
3385           something is wrong until it gets to the "}" and encounters the
3386           missing (or unexpected) comma.  The syntax error will be reported
3387           close to the "}", but you'll need to change something near the "{"
3388           such as using a unary "+" or semicolon to give Perl some help:
3389
3390            my %hash = map {  "\L$_" => 1  } @array # perl guesses EXPR. wrong
3391            my %hash = map { +"\L$_" => 1  } @array # perl guesses BLOCK. right
3392            my %hash = map {; "\L$_" => 1  } @array # this also works
3393            my %hash = map { ("\L$_" => 1) } @array # as does this
3394            my %hash = map {  lc($_) => 1  } @array # and this.
3395            my %hash = map +( lc($_) => 1 ), @array # this is EXPR and works!
3396
3397            my %hash = map  ( lc($_), 1 ),   @array # evaluates to (1, @array)
3398
3399           or to force an anon hash constructor use "+{":
3400
3401               my @hashes = map +{ lc($_) => 1 }, @array # EXPR, so needs
3402                                                         # comma at end
3403
3404           to get a list of anonymous hashes each with only one entry apiece.
3405
3406       mkdir FILENAME,MODE
3407       mkdir FILENAME
3408       mkdir
3409           Creates the directory specified by FILENAME, with permissions
3410           specified by MODE (as modified by "umask").  If it succeeds it
3411           returns true; otherwise it returns false and sets $! (errno).  MODE
3412           defaults to 0777 if omitted, and FILENAME defaults to $_ if
3413           omitted.
3414
3415           In general, it is better to create directories with a permissive
3416           MODE and let the user modify that with their "umask" than it is to
3417           supply a restrictive MODE and give the user no way to be more
3418           permissive.  The exceptions to this rule are when the file or
3419           directory should be kept private (mail files, for instance).  The
3420           documentation for "umask" discusses the choice of MODE in more
3421           detail.
3422
3423           Note that according to the POSIX 1003.1-1996 the FILENAME may have
3424           any number of trailing slashes.  Some operating and filesystems do
3425           not get this right, so Perl automatically removes all trailing
3426           slashes to keep everyone happy.
3427
3428           To recursively create a directory structure, look at the
3429           "make_path" function of the File::Path module.
3430
3431       msgctl ID,CMD,ARG
3432           Calls the System V IPC function msgctl(2).  You'll probably have to
3433           say
3434
3435               use IPC::SysV;
3436
3437           first to get the correct constant definitions.  If CMD is
3438           "IPC_STAT", then ARG must be a variable that will hold the returned
3439           "msqid_ds" structure.  Returns like "ioctl": the undefined value
3440           for error, "0 but true" for zero, or the actual return value
3441           otherwise.  See also "SysV IPC" in perlipc and the documentation
3442           for "IPC::SysV" and "IPC::Semaphore".
3443
3444           Portability issues: "msgctl" in perlport.
3445
3446       msgget KEY,FLAGS
3447           Calls the System V IPC function msgget(2).  Returns the message
3448           queue id, or "undef" on error.  See also "SysV IPC" in perlipc and
3449           the documentation for "IPC::SysV" and "IPC::Msg".
3450
3451           Portability issues: "msgget" in perlport.
3452
3453       msgrcv ID,VAR,SIZE,TYPE,FLAGS
3454           Calls the System V IPC function msgrcv to receive a message from
3455           message queue ID into variable VAR with a maximum message size of
3456           SIZE.  Note that when a message is received, the message type as a
3457           native long integer will be the first thing in VAR, followed by the
3458           actual message.  This packing may be opened with "unpack("l! a*")".
3459           Taints the variable.  Returns true if successful, false on error.
3460           See also "SysV IPC" in perlipc and the documentation for
3461           "IPC::SysV" and "IPC::Msg".
3462
3463           Portability issues: "msgrcv" in perlport.
3464
3465       msgsnd ID,MSG,FLAGS
3466           Calls the System V IPC function msgsnd to send the message MSG to
3467           the message queue ID.  MSG must begin with the native long integer
3468           message type, followed by the message itself.  This kind of packing
3469           can be achieved with "pack("l! a*", $type, $message)".  Returns
3470           true if successful, false on error.  See also "SysV IPC" in perlipc
3471           and the documentation for "IPC::SysV" and "IPC::Msg".
3472
3473           Portability issues: "msgsnd" in perlport.
3474
3475       my VARLIST
3476       my TYPE VARLIST
3477       my VARLIST : ATTRS
3478       my TYPE VARLIST : ATTRS
3479           A "my" declares the listed variables to be local (lexically) to the
3480           enclosing block, file, or "eval".  If more than one variable is
3481           listed, the list must be placed in parentheses.
3482
3483           Note that with a parenthesised list, "undef" can be used as a dummy
3484           placeholder, for example to skip assignment of initial values:
3485
3486               my ( undef, $min, $hour ) = localtime;
3487
3488           Redeclaring a variable in the same scope or statement will "shadow"
3489           the previous declaration, creating a new instance and preventing
3490           access to the previous one. This is usually undesired and, if
3491           warnings are enabled, will result in a warning in the "shadow"
3492           category.
3493
3494           The exact semantics and interface of TYPE and ATTRS are still
3495           evolving.  TYPE may be a bareword, a constant declared with "use
3496           constant", or "__PACKAGE__".  It is currently bound to the use of
3497           the fields pragma, and attributes are handled using the attributes
3498           pragma, or starting from Perl 5.8.0 also via the
3499           Attribute::Handlers module.  See "Private Variables via my()" in
3500           perlsub for details.
3501
3502       next LABEL
3503       next EXPR
3504       next
3505           The "next" command is like the "continue" statement in C; it starts
3506           the next iteration of the loop:
3507
3508               LINE: while (<STDIN>) {
3509                   next LINE if /^#/;  # discard comments
3510                   #...
3511               }
3512
3513           Note that if there were a "continue" block on the above, it would
3514           get executed even on discarded lines.  If LABEL is omitted, the
3515           command refers to the innermost enclosing loop.  The "next EXPR"
3516           form, available as of Perl 5.18.0, allows a label name to be
3517           computed at run time, being otherwise identical to "next LABEL".
3518
3519           "next" cannot return a value from a block that typically returns a
3520           value, such as "eval {}", "sub {}", or "do {}". It will perform its
3521           flow control behavior, which precludes any return value. It should
3522           not be used to exit a "grep" or "map" operation.
3523
3524           Note that a block by itself is semantically identical to a loop
3525           that executes once.  Thus "next" will exit such a block early.
3526
3527           See also "continue" for an illustration of how "last", "next", and
3528           "redo" work.
3529
3530           Unlike most named operators, this has the same precedence as
3531           assignment.  It is also exempt from the looks-like-a-function rule,
3532           so "next ("foo")."bar"" will cause "bar" to be part of the argument
3533           to "next".
3534
3535       no MODULE VERSION LIST
3536       no MODULE VERSION
3537       no MODULE LIST
3538       no MODULE
3539       no VERSION
3540           See the "use" function, of which "no" is the opposite.
3541
3542       oct EXPR
3543       oct Interprets EXPR as an octal string and returns the corresponding
3544           value.  An octal string consists of octal digits and, as of Perl
3545           5.33.5, an optional "0o" or "o" prefix.  Each octal digit may be
3546           preceded by a single underscore, which will be ignored.  (If EXPR
3547           happens to start off with "0x" or "x", interprets it as a hex
3548           string.  If EXPR starts off with "0b" or "b", it is interpreted as
3549           a binary string.  Leading whitespace is ignored in all three
3550           cases.)  The following will handle decimal, binary, octal, and hex
3551           in standard Perl notation:
3552
3553               $val = oct($val) if $val =~ /^0/;
3554
3555           If EXPR is omitted, uses $_.   To go the other way (produce a
3556           number in octal), use "sprintf" or "printf":
3557
3558               my $dec_perms = (stat("filename"))[2] & 07777;
3559               my $oct_perm_str = sprintf "%o", $perms;
3560
3561           The "oct" function is commonly used when a string such as 644 needs
3562           to be converted into a file mode, for example.  Although Perl
3563           automatically converts strings into numbers as needed, this
3564           automatic conversion assumes base 10.
3565
3566           Leading white space is ignored without warning, as too are any
3567           trailing non-digits, such as a decimal point ("oct" only handles
3568           non-negative integers, not negative integers or floating point).
3569
3570       open FILEHANDLE,MODE,EXPR
3571       open FILEHANDLE,MODE,EXPR,LIST
3572       open FILEHANDLE,MODE,REFERENCE
3573       open FILEHANDLE,EXPR
3574       open FILEHANDLE
3575           Associates an internal FILEHANDLE with the external file specified
3576           by EXPR. That filehandle will subsequently allow you to perform I/O
3577           operations on that file, such as reading from it or writing to it.
3578
3579           Instead of a filename, you may specify an external command (plus an
3580           optional argument list) or a scalar reference, in order to open
3581           filehandles on commands or in-memory scalars, respectively.
3582
3583           A thorough reference to "open" follows. For a gentler introduction
3584           to the basics of "open", see also the perlopentut manual page.
3585
3586           Working with files
3587               Most often, "open" gets invoked with three arguments: the
3588               required FILEHANDLE (usually an empty scalar variable),
3589               followed by MODE (usually a literal describing the I/O mode the
3590               filehandle will use), and then the filename  that the new
3591               filehandle will refer to.
3592
3593               Simple examples
3594                   Reading from a file:
3595
3596                       open(my $fh, "<", "input.txt")
3597                           or die "Can't open < input.txt: $!";
3598
3599                       # Process every line in input.txt
3600                       while (my $line = <$fh>) {
3601                           #
3602                           # ... do something interesting with $line here ...
3603                           #
3604                       }
3605
3606                   or writing to one:
3607
3608                       open(my $fh, ">", "output.txt")
3609                           or die "Can't open > output.txt: $!";
3610
3611                       print $fh "This line gets printed into output.txt.\n";
3612
3613                   For a summary of common filehandle operations such as
3614                   these, see "Files and I/O" in perlintro.
3615
3616               About filehandles
3617                   The first argument to "open", labeled FILEHANDLE in this
3618                   reference, is usually a scalar variable. (Exceptions exist,
3619                   described in "Other considerations", below.) If the call to
3620                   "open" succeeds, then the expression provided as FILEHANDLE
3621                   will get assigned an open filehandle. That filehandle
3622                   provides an internal reference to the specified external
3623                   file, conveniently stored in a Perl variable, and ready for
3624                   I/O operations such as reading and writing.
3625
3626               About modes
3627                   When calling "open" with three or more arguments, the
3628                   second argument -- labeled MODE here -- defines the open
3629                   mode. MODE is usually a literal string comprising special
3630                   characters that define the intended I/O role of the
3631                   filehandle being created: whether it's read-only, or read-
3632                   and-write, and so on.
3633
3634                   If MODE is "<", the file is opened for input (read-only).
3635                   If MODE is ">", the file is opened for output, with
3636                   existing files first being truncated ("clobbered") and
3637                   nonexisting files newly created.  If MODE is ">>", the file
3638                   is opened for appending, again being created if necessary.
3639
3640                   You can put a "+" in front of the ">" or "<" to indicate
3641                   that you want both read and write access to the file; thus
3642                   "+<" is almost always preferred for read/write updates--the
3643                   "+>" mode would clobber the file first.  You can't usually
3644                   use either read-write mode for updating textfiles, since
3645                   they have variable-length records.  See the -i switch in
3646                   perlrun for a better approach.  The file is created with
3647                   permissions of 0666 modified by the process's "umask"
3648                   value.
3649
3650                   These various prefixes correspond to the fopen(3) modes of
3651                   "r", "r+", "w", "w+", "a", and "a+".
3652
3653                   More examples of different modes in action:
3654
3655                    # Open a file for concatenation
3656                    open(my $log, ">>", "/usr/spool/news/twitlog")
3657                        or warn "Couldn't open log file; discarding input";
3658
3659                    # Open a file for reading and writing
3660                    open(my $dbase, "+<", "dbase.mine")
3661                        or die "Can't open 'dbase.mine' for update: $!";
3662
3663               Checking the return value
3664                   Open returns nonzero on success, the undefined value
3665                   otherwise.  If the "open" involved a pipe, the return value
3666                   happens to be the pid of the subprocess.
3667
3668                   When opening a file, it's seldom a good idea to continue if
3669                   the request failed, so "open" is frequently used with
3670                   "die". Even if you want your code to do something other
3671                   than "die" on a failed open, you should still always check
3672                   the return value from opening a file.
3673
3674           Specifying I/O layers in MODE
3675               You can use the three-argument form of open to specify I/O
3676               layers (sometimes referred to as "disciplines") to apply to the
3677               new filehandle. These affect how the input and output are
3678               processed (see open and PerlIO for more details).  For example:
3679
3680                   # loads PerlIO::encoding automatically
3681                   open(my $fh, "<:encoding(UTF-8)", $filename)
3682                       || die "Can't open UTF-8 encoded $filename: $!";
3683
3684               This opens the UTF8-encoded file containing Unicode characters;
3685               see perluniintro.  Note that if layers are specified in the
3686               three-argument form, then default layers stored in "${^OPEN}"
3687               (usually set by the open pragma or the switch "-CioD") are
3688               ignored.  Those layers will also be ignored if you specify a
3689               colon with no name following it.  In that case the default
3690               layer for the operating system (:raw on Unix, :crlf on Windows)
3691               is used.
3692
3693               On some systems (in general, DOS- and Windows-based systems)
3694               "binmode" is necessary when you're not working with a text
3695               file.  For the sake of portability it is a good idea always to
3696               use it when appropriate, and never to use it when it isn't
3697               appropriate.  Also, people can set their I/O to be by default
3698               UTF8-encoded Unicode, not bytes.
3699
3700           Using "undef" for temporary files
3701               As a special case the three-argument form with a read/write
3702               mode and the third argument being "undef":
3703
3704                   open(my $tmp, "+>", undef) or die ...
3705
3706               opens a filehandle to a newly created empty anonymous temporary
3707               file.  (This happens under any mode, which makes "+>" the only
3708               useful and sensible mode to use.)  You will need to "seek" to
3709               do the reading.
3710
3711           Opening a filehandle into an in-memory scalar
3712               You can open filehandles directly to Perl scalars instead of a
3713               file or other resource external to the program. To do so,
3714               provide a reference to that scalar as the third argument to
3715               "open", like so:
3716
3717                open(my $memory, ">", \$var)
3718                    or die "Can't open memory file: $!";
3719                print $memory "foo!\n";    # output will appear in $var
3720
3721               To (re)open "STDOUT" or "STDERR" as an in-memory file, close it
3722               first:
3723
3724                   close STDOUT;
3725                   open(STDOUT, ">", \$variable)
3726                       or die "Can't open STDOUT: $!";
3727
3728               The scalars for in-memory files are treated as octet strings:
3729               unless the file is being opened with truncation the scalar may
3730               not contain any code points over 0xFF.
3731
3732               Opening in-memory files can fail for a variety of reasons.  As
3733               with any other "open", check the return value for success.
3734
3735               Technical note: This feature works only when Perl is built with
3736               PerlIO -- the default, except with older (pre-5.16) Perl
3737               installations that were configured to not include it (e.g. via
3738               "Configure -Uuseperlio"). You can see whether your Perl was
3739               built with PerlIO by running "perl -V:useperlio".  If it says
3740               'define', you have PerlIO; otherwise you don't.
3741
3742               See perliol for detailed info on PerlIO.
3743
3744           Opening a filehandle into a command
3745               If MODE is "|-", then the filename is interpreted as a command
3746               to which output is to be piped, and if MODE is "-|", the
3747               filename is interpreted as a command that pipes output to us.
3748               In the two-argument (and one-argument) form, one should replace
3749               dash ("-") with the command.  See "Using open() for IPC" in
3750               perlipc for more examples of this.  (You are not allowed to
3751               "open" to a command that pipes both in and out, but see
3752               IPC::Open2, IPC::Open3, and "Bidirectional Communication with
3753               Another Process" in perlipc for alternatives.)
3754
3755                open(my $article_fh, "-|", "caesar <$article")  # decrypt
3756                                                                # article
3757                    or die "Can't start caesar: $!";
3758
3759                open(my $article_fh, "caesar <$article |")      # ditto
3760                    or die "Can't start caesar: $!";
3761
3762                open(my $out_fh, "|-", "sort >Tmp$$")    # $$ is our process id
3763                    or die "Can't start sort: $!";
3764
3765               In the form of pipe opens taking three or more arguments, if
3766               LIST is specified (extra arguments after the command name) then
3767               LIST becomes arguments to the command invoked if the platform
3768               supports it.  The meaning of "open" with more than three
3769               arguments for non-pipe modes is not yet defined, but
3770               experimental "layers" may give extra LIST arguments meaning.
3771
3772               If you open a pipe on the command "-" (that is, specify either
3773               "|-" or "-|" with the one- or two-argument forms of "open"), an
3774               implicit "fork" is done, so "open" returns twice: in the parent
3775               process it returns the pid of the child process, and in the
3776               child process it returns (a defined) 0.  Use "defined($pid)" or
3777               "//" to determine whether the open was successful.
3778
3779               For example, use either
3780
3781                  my $child_pid = open(my $from_kid, "-|")
3782                       // die "Can't fork: $!";
3783
3784               or
3785
3786                  my $child_pid = open(my $to_kid,   "|-")
3787                       // die "Can't fork: $!";
3788
3789               followed by
3790
3791                   if ($child_pid) {
3792                       # am the parent:
3793                       # either write $to_kid or else read $from_kid
3794                       ...
3795                      waitpid $child_pid, 0;
3796                   } else {
3797                       # am the child; use STDIN/STDOUT normally
3798                       ...
3799                       exit;
3800                   }
3801
3802               The filehandle behaves normally for the parent, but I/O to that
3803               filehandle is piped from/to the STDOUT/STDIN of the child
3804               process.  In the child process, the filehandle isn't
3805               opened--I/O happens from/to the new STDOUT/STDIN.  Typically
3806               this is used like the normal piped open when you want to
3807               exercise more control over just how the pipe command gets
3808               executed, such as when running setuid and you don't want to
3809               have to scan shell commands for metacharacters.
3810
3811               The following blocks are more or less equivalent:
3812
3813                   open(my $fh, "|tr '[a-z]' '[A-Z]'");
3814                   open(my $fh, "|-", "tr '[a-z]' '[A-Z]'");
3815                   open(my $fh, "|-") || exec 'tr', '[a-z]', '[A-Z]';
3816                   open(my $fh, "|-", "tr", '[a-z]', '[A-Z]');
3817
3818                   open(my $fh, "cat -n '$file'|");
3819                   open(my $fh, "-|", "cat -n '$file'");
3820                   open(my $fh, "-|") || exec "cat", "-n", $file;
3821                   open(my $fh, "-|", "cat", "-n", $file);
3822
3823               The last two examples in each block show the pipe as "list
3824               form", which is not yet supported on all platforms. (If your
3825               platform has a real "fork", such as Linux and macOS, you can
3826               use the list form; it also works on Windows with Perl 5.22 or
3827               later.) You would want to use the list form of the pipe so you
3828               can pass literal arguments to the command without risk of the
3829               shell interpreting any shell metacharacters in them. However,
3830               this also bars you from opening pipes to commands that
3831               intentionally contain shell metacharacters, such as:
3832
3833                   open(my $fh, "|cat -n | expand -4 | lpr")
3834                       || die "Can't open pipeline to lpr: $!";
3835
3836               See "Safe Pipe Opens" in perlipc for more examples of this.
3837
3838           Duping filehandles
3839               You may also, in the Bourne shell tradition, specify an EXPR
3840               beginning with ">&", in which case the rest of the string is
3841               interpreted as the name of a filehandle (or file descriptor, if
3842               numeric) to be duped (as in dup(2)) and opened.  You may use
3843               "&" after ">", ">>", "<", "+>", "+>>", and "+<".  The mode you
3844               specify should match the mode of the original filehandle.
3845               (Duping a filehandle does not take into account any existing
3846               contents of IO buffers.)  If you use the three-argument form,
3847               then you can pass either a number, the name of a filehandle, or
3848               the normal "reference to a glob".
3849
3850               Here is a script that saves, redirects, and restores "STDOUT"
3851               and "STDERR" using various methods:
3852
3853                   #!/usr/bin/perl
3854                   open(my $oldout, ">&STDOUT")
3855                       or die "Can't dup STDOUT: $!";
3856                   open(OLDERR,     ">&", \*STDERR)
3857                       or die "Can't dup STDERR: $!";
3858
3859                   open(STDOUT, '>', "foo.out")
3860                       or die "Can't redirect STDOUT: $!";
3861                   open(STDERR, ">&STDOUT")
3862                       or die "Can't dup STDOUT: $!";
3863
3864                   select STDERR; $| = 1;  # make unbuffered
3865                   select STDOUT; $| = 1;  # make unbuffered
3866
3867                   print STDOUT "stdout 1\n";  # this works for
3868                   print STDERR "stderr 1\n";  # subprocesses too
3869
3870                   open(STDOUT, ">&", $oldout)
3871                       or die "Can't dup \$oldout: $!";
3872                   open(STDERR, ">&OLDERR")
3873                       or die "Can't dup OLDERR: $!";
3874
3875                   print STDOUT "stdout 2\n";
3876                   print STDERR "stderr 2\n";
3877
3878               If you specify '<&=X', where "X" is a file descriptor number or
3879               a filehandle, then Perl will do an equivalent of C's fdopen(3)
3880               of that file descriptor (and not call dup(2)); this is more
3881               parsimonious of file descriptors.  For example:
3882
3883                   # open for input, reusing the fileno of $fd
3884                   open(my $fh, "<&=", $fd)
3885
3886               or
3887
3888                   open(my $fh, "<&=$fd")
3889
3890               or
3891
3892                   # open for append, using the fileno of $oldfh
3893                   open(my $fh, ">>&=", $oldfh)
3894
3895               Being parsimonious on filehandles is also useful (besides being
3896               parsimonious) for example when something is dependent on file
3897               descriptors, like for example locking using "flock".  If you do
3898               just "open(my $A, ">>&", $B)", the filehandle $A will not have
3899               the same file descriptor as $B, and therefore "flock($A)" will
3900               not "flock($B)" nor vice versa.  But with "open(my $A, ">>&=",
3901               $B)", the filehandles will share the same underlying system
3902               file descriptor.
3903
3904               Note that under Perls older than 5.8.0, Perl uses the standard
3905               C library's' fdopen(3) to implement the "=" functionality.  On
3906               many Unix systems, fdopen(3) fails when file descriptors exceed
3907               a certain value, typically 255.  For Perls 5.8.0 and later,
3908               PerlIO is (most often) the default.
3909
3910           Legacy usage
3911               This section describes ways to call "open" outside of best
3912               practices; you may encounter these uses in older code. Perl
3913               does not consider their use deprecated, exactly, but neither is
3914               it recommended in new code, for the sake of clarity and
3915               readability.
3916
3917               Specifying mode and filename as a single argument
3918                   In the one- and two-argument forms of the call, the mode
3919                   and filename should be concatenated (in that order),
3920                   preferably separated by white space.  You can--but
3921                   shouldn't--omit the mode in these forms when that mode is
3922                   "<".  It is safe to use the two-argument form of "open" if
3923                   the filename argument is a known literal.
3924
3925                    open(my $dbase, "+<dbase.mine")          # ditto
3926                        or die "Can't open 'dbase.mine' for update: $!";
3927
3928                   In the two-argument (and one-argument) form, opening "<-"
3929                   or "-" opens STDIN and opening ">-" opens STDOUT.
3930
3931                   New code should favor the three-argument form of "open"
3932                   over this older form. Declaring the mode and the filename
3933                   as two distinct arguments avoids any confusion between the
3934                   two.
3935
3936               Calling "open" with one argument via global variables
3937                   As a shortcut, a one-argument call takes the filename from
3938                   the global scalar variable of the same name as the
3939                   filehandle:
3940
3941                       $ARTICLE = 100;
3942                       open(ARTICLE)
3943                           or die "Can't find article $ARTICLE: $!\n";
3944
3945                   Here $ARTICLE must be a global (package) scalar variable -
3946                   not one declared with "my" or "state".
3947
3948               Assigning a filehandle to a bareword
3949                   An older style is to use a bareword as the filehandle, as
3950
3951                       open(FH, "<", "input.txt")
3952                          or die "Can't open < input.txt: $!";
3953
3954                   Then you can use "FH" as the filehandle, in "close FH" and
3955                   "<FH>" and so on.  Note that it's a global variable, so
3956                   this form is not recommended when dealing with filehandles
3957                   other than Perl's built-in ones (e.g. STDOUT and STDIN).
3958                   In fact, using a bareword for the filehandle is an error
3959                   when the "bareword_filehandles" feature has been disabled.
3960                   This feature is disabled by default when in the scope of
3961                   "use v5.36.0" or later.
3962
3963           Other considerations
3964               Automatic filehandle closure
3965                   The filehandle will be closed when its reference count
3966                   reaches zero. If it is a lexically scoped variable declared
3967                   with "my", that usually means the end of the enclosing
3968                   scope.  However, this automatic close does not check for
3969                   errors, so it is better to explicitly close filehandles,
3970                   especially those used for writing:
3971
3972                       close($handle)
3973                          || warn "close failed: $!";
3974
3975               Automatic pipe flushing
3976                   Perl will attempt to flush all files opened for output
3977                   before any operation that may do a fork, but this may not
3978                   be supported on some platforms (see perlport).  To be safe,
3979                   you may need to set $| ($AUTOFLUSH in English) or call the
3980                   "autoflush" method of "IO::Handle" on any open handles.
3981
3982                   On systems that support a close-on-exec flag on files, the
3983                   flag will be set for the newly opened file descriptor as
3984                   determined by the value of $^F.  See "$^F" in perlvar.
3985
3986                   Closing any piped filehandle causes the parent process to
3987                   wait for the child to finish, then returns the status value
3988                   in $? and "${^CHILD_ERROR_NATIVE}".
3989
3990               Direct versus by-reference assignment of filehandles
3991                   If FILEHANDLE -- the first argument in a call to "open" --
3992                   is an undefined scalar variable (or array or hash element),
3993                   a new filehandle is autovivified, meaning that the variable
3994                   is assigned a reference to a newly allocated anonymous
3995                   filehandle.  Otherwise if FILEHANDLE is an expression, its
3996                   value is the real filehandle.  (This is considered a
3997                   symbolic reference, so "use strict "refs"" should not be in
3998                   effect.)
3999
4000               Whitespace and special characters in the filename argument
4001                   The filename passed to the one- and two-argument forms of
4002                   "open" will have leading and trailing whitespace deleted
4003                   and normal redirection characters honored.  This property,
4004                   known as "magic open", can often be used to good effect.  A
4005                   user could specify a filename of "rsh cat file |", or you
4006                   could change certain filenames as needed:
4007
4008                       $filename =~ s/(.*\.gz)\s*$/gzip -dc < $1|/;
4009                       open(my $fh, $filename)
4010                           or die "Can't open $filename: $!";
4011
4012                   Use the three-argument form to open a file with arbitrary
4013                   weird characters in it,
4014
4015                       open(my $fh, "<", $file)
4016                           || die "Can't open $file: $!";
4017
4018                   otherwise it's necessary to protect any leading and
4019                   trailing whitespace:
4020
4021                       $file =~ s#^(\s)#./$1#;
4022                       open(my $fh, "< $file\0")
4023                           || die "Can't open $file: $!";
4024
4025                   (this may not work on some bizarre filesystems).  One
4026                   should conscientiously choose between the magic and three-
4027                   argument form of "open":
4028
4029                       open(my $in, $ARGV[0]) || die "Can't open $ARGV[0]: $!";
4030
4031                   will allow the user to specify an argument of the form "rsh
4032                   cat file |", but will not work on a filename that happens
4033                   to have a trailing space, while
4034
4035                       open(my $in, "<", $ARGV[0])
4036                           || die "Can't open $ARGV[0]: $!";
4037
4038                   will have exactly the opposite restrictions. (However, some
4039                   shells support the syntax "perl your_program.pl <( rsh cat
4040                   file )", which produces a filename that can be opened
4041                   normally.)
4042
4043               Invoking C-style "open"
4044                   If you want a "real" C open(2), then you should use the
4045                   "sysopen" function, which involves no such magic (but uses
4046                   different filemodes than Perl "open", which corresponds to
4047                   C fopen(3)).  This is another way to protect your filenames
4048                   from interpretation.  For example:
4049
4050                       use IO::Handle;
4051                       sysopen(my $fh, $path, O_RDWR|O_CREAT|O_EXCL)
4052                           or die "Can't open $path: $!";
4053                       $fh->autoflush(1);
4054                       print $fh "stuff $$\n";
4055                       seek($fh, 0, 0);
4056                       print "File contains: ", readline($fh);
4057
4058                   See "seek" for some details about mixing reading and
4059                   writing.
4060
4061               Portability issues
4062                   See "open" in perlport.
4063
4064       opendir DIRHANDLE,EXPR
4065           Opens a directory named EXPR for processing by "readdir",
4066           "telldir", "seekdir", "rewinddir", and "closedir".  Returns true if
4067           successful.  DIRHANDLE may be an expression whose value can be used
4068           as an indirect dirhandle, usually the real dirhandle name.  If
4069           DIRHANDLE is an undefined scalar variable (or array or hash
4070           element), the variable is assigned a reference to a new anonymous
4071           dirhandle; that is, it's autovivified.  Dirhandles are the same
4072           objects as filehandles; an I/O object can only be open as one of
4073           these handle types at once.
4074
4075           See the example at "readdir".
4076
4077       ord EXPR
4078       ord Returns the numeric value of the first character of EXPR.  If EXPR
4079           is an empty string, returns 0.  If EXPR is omitted, uses $_.  (Note
4080           character, not byte.)
4081
4082           For the reverse, see "chr".  See perlunicode for more about
4083           Unicode.
4084
4085       our VARLIST
4086       our TYPE VARLIST
4087       our VARLIST : ATTRS
4088       our TYPE VARLIST : ATTRS
4089           "our" makes a lexical alias to a package (i.e. global) variable of
4090           the same name in the current package for use within the current
4091           lexical scope.
4092
4093           "our" has the same scoping rules as "my" or "state", meaning that
4094           it is only valid within a lexical scope.  Unlike "my" and "state",
4095           which both declare new (lexical) variables, "our" only creates an
4096           alias to an existing variable: a package variable of the same name.
4097
4098           This means that when "use strict 'vars'" is in effect, "our" lets
4099           you use a package variable without qualifying it with the package
4100           name, but only within the lexical scope of the "our" declaration.
4101           This applies immediately--even within the same statement.
4102
4103               package Foo;
4104               use v5.36;  # which implies "use strict;"
4105
4106               $Foo::foo = 23;
4107
4108               {
4109                   our $foo;   # alias to $Foo::foo
4110                   print $foo; # prints 23
4111               }
4112
4113               print $Foo::foo; # prints 23
4114
4115               print $foo; # ERROR: requires explicit package name
4116
4117           This works even if the package variable has not been used before,
4118           as package variables spring into existence when first used.
4119
4120               package Foo;
4121               use v5.36;
4122
4123               our $foo = 23;   # just like $Foo::foo = 23
4124
4125               print $Foo::foo; # prints 23
4126
4127           Because the variable becomes legal immediately under "use strict
4128           'vars'", so long as there is no variable with that name is already
4129           in scope, you can then reference the package variable again even
4130           within the same statement.
4131
4132               package Foo;
4133               use v5.36;
4134
4135               my  $foo = $foo; # error, undeclared $foo on right-hand side
4136               our $foo = $foo; # no errors
4137
4138           If more than one variable is listed, the list must be placed in
4139           parentheses.
4140
4141               our($bar, $baz);
4142
4143           An "our" declaration declares an alias for a package variable that
4144           will be visible across its entire lexical scope, even across
4145           package boundaries.  The package in which the variable is entered
4146           is determined at the point of the declaration, not at the point of
4147           use.  This means the following behavior holds:
4148
4149               package Foo;
4150               our $bar;      # declares $Foo::bar for rest of lexical scope
4151               $bar = 20;
4152
4153               package Bar;
4154               print $bar;    # prints 20, as it refers to $Foo::bar
4155
4156           Multiple "our" declarations with the same name in the same lexical
4157           scope are allowed if they are in different packages.  If they
4158           happen to be in the same package, Perl will emit warnings if you
4159           have asked for them, just like multiple "my" declarations.  Unlike
4160           a second "my" declaration, which will bind the name to a fresh
4161           variable, a second "our" declaration in the same package, in the
4162           same scope, is merely redundant.
4163
4164               use warnings;
4165               package Foo;
4166               our $bar;      # declares $Foo::bar for rest of lexical scope
4167               $bar = 20;
4168
4169               package Bar;
4170               our $bar = 30; # declares $Bar::bar for rest of lexical scope
4171               print $bar;    # prints 30
4172
4173               our $bar;      # emits warning but has no other effect
4174               print $bar;    # still prints 30
4175
4176           An "our" declaration may also have a list of attributes associated
4177           with it.
4178
4179           The exact semantics and interface of TYPE and ATTRS are still
4180           evolving.  TYPE is currently bound to the use of the fields pragma,
4181           and attributes are handled using the attributes pragma, or,
4182           starting from Perl 5.8.0, also via the Attribute::Handlers module.
4183           See "Private Variables via my()" in perlsub for details.
4184
4185           Note that with a parenthesised list, "undef" can be used as a dummy
4186           placeholder, for example to skip assignment of initial values:
4187
4188               our ( undef, $min, $hour ) = localtime;
4189
4190           "our" differs from "use vars", which allows use of an unqualified
4191           name only within the affected package, but across scopes.
4192
4193       pack TEMPLATE,LIST
4194           Takes a LIST of values and converts it into a string using the
4195           rules given by the TEMPLATE.  The resulting string is the
4196           concatenation of the converted values.  Typically, each converted
4197           value looks like its machine-level representation.  For example, on
4198           32-bit machines an integer may be represented by a sequence of 4
4199           bytes, which  will in Perl be presented as a string that's 4
4200           characters long.
4201
4202           See perlpacktut for an introduction to this function.
4203
4204           The TEMPLATE is a sequence of characters that give the order and
4205           type of values, as follows:
4206
4207               a  A string with arbitrary binary data, will be null padded.
4208               A  A text (ASCII) string, will be space padded.
4209               Z  A null-terminated (ASCIZ) string, will be null padded.
4210
4211               b  A bit string (ascending bit order inside each byte,
4212                  like vec()).
4213               B  A bit string (descending bit order inside each byte).
4214               h  A hex string (low nybble first).
4215               H  A hex string (high nybble first).
4216
4217               c  A signed char (8-bit) value.
4218               C  An unsigned char (octet) value.
4219               W  An unsigned char value (can be greater than 255).
4220
4221               s  A signed short (16-bit) value.
4222               S  An unsigned short value.
4223
4224               l  A signed long (32-bit) value.
4225               L  An unsigned long value.
4226
4227               q  A signed quad (64-bit) value.
4228               Q  An unsigned quad value.
4229                    (Quads are available only if your system supports 64-bit
4230                     integer values _and_ if Perl has been compiled to support
4231                     those.  Raises an exception otherwise.)
4232
4233               i  A signed integer value.
4234               I  An unsigned integer value.
4235                    (This 'integer' is _at_least_ 32 bits wide.  Its exact
4236                     size depends on what a local C compiler calls 'int'.)
4237
4238               n  An unsigned short (16-bit) in "network" (big-endian) order.
4239               N  An unsigned long (32-bit) in "network" (big-endian) order.
4240               v  An unsigned short (16-bit) in "VAX" (little-endian) order.
4241               V  An unsigned long (32-bit) in "VAX" (little-endian) order.
4242
4243               j  A Perl internal signed integer value (IV).
4244               J  A Perl internal unsigned integer value (UV).
4245
4246               f  A single-precision float in native format.
4247               d  A double-precision float in native format.
4248
4249               F  A Perl internal floating-point value (NV) in native format
4250               D  A float of long-double precision in native format.
4251                    (Long doubles are available only if your system supports
4252                     long double values. Raises an exception otherwise.
4253                     Note that there are different long double formats.)
4254
4255               p  A pointer to a null-terminated string.
4256               P  A pointer to a structure (fixed-length string).
4257
4258               u  A uuencoded string.
4259               U  A Unicode character number.  Encodes to a character in char-
4260                  acter mode and UTF-8 (or UTF-EBCDIC in EBCDIC platforms) in
4261                  byte mode.  Also on EBCDIC platforms, the character number will
4262                  be the native EBCDIC value for character numbers below 256.
4263                  This allows most programs using this feature to not have to
4264                  care which type of platform they are running on.
4265
4266               w  A BER compressed integer (not an ASN.1 BER, see perlpacktut
4267                  for details).  Its bytes represent an unsigned integer in
4268                  base 128, most significant digit first, with as few digits
4269                  as possible.  Bit eight (the high bit) is set on each byte
4270                  except the last.
4271
4272               x  A null byte (a.k.a ASCII NUL, "\000", chr(0))
4273               X  Back up a byte.
4274               @  Null-fill or truncate to absolute position, counted from the
4275                  start of the innermost ()-group.
4276               .  Null-fill or truncate to absolute position specified by
4277                  the value.
4278               (  Start of a ()-group.
4279
4280           One or more modifiers below may optionally follow certain letters
4281           in the TEMPLATE (the second column lists letters for which the
4282           modifier is valid):
4283
4284               !   sSlLiI     Forces native (short, long, int) sizes instead
4285                              of fixed (16-/32-bit) sizes.
4286
4287               !   xX         Make x and X act as alignment commands.
4288
4289               !   nNvV       Treat integers as signed instead of unsigned.
4290
4291               !   @.         Specify position as byte offset in the internal
4292                              representation of the packed string.  Efficient
4293                              but dangerous.
4294
4295               >   sSiIlLqQ   Force big-endian byte-order on the type.
4296                   jJfFdDpP   (The "big end" touches the construct.)
4297
4298               <   sSiIlLqQ   Force little-endian byte-order on the type.
4299                   jJfFdDpP   (The "little end" touches the construct.)
4300
4301           The ">" and "<" modifiers can also be used on "()" groups to force
4302           a particular byte-order on all components in that group, including
4303           all its subgroups.
4304
4305           The following rules apply:
4306
4307           •   Each letter may optionally be followed by a number indicating
4308               the repeat count.  A numeric repeat count may optionally be
4309               enclosed in brackets, as in "pack("C[80]", @arr)".  The repeat
4310               count gobbles that many values from the LIST when used with all
4311               format types other than "a", "A", "Z", "b", "B", "h", "H", "@",
4312               ".", "x", "X", and "P", where it means something else,
4313               described below.  Supplying a "*" for the repeat count instead
4314               of a number means to use however many items are left, except
4315               for:
4316
4317               •   "@", "x", and "X", where it is equivalent to 0.
4318
4319               •   <.>, where it means relative to the start of the string.
4320
4321               •   "u", where it is equivalent to 1 (or 45, which here is
4322                   equivalent).
4323
4324               One can replace a numeric repeat count with a template letter
4325               enclosed in brackets to use the packed byte length of the
4326               bracketed template for the repeat count.
4327
4328               For example, the template "x[L]" skips as many bytes as in a
4329               packed long, and the template "$t X[$t] $t" unpacks twice
4330               whatever $t (when variable-expanded) unpacks.  If the template
4331               in brackets contains alignment commands (such as "x![d]"), its
4332               packed length is calculated as if the start of the template had
4333               the maximal possible alignment.
4334
4335               When used with "Z", a "*" as the repeat count is guaranteed to
4336               add a trailing null byte, so the resulting string is always one
4337               byte longer than the byte length of the item itself.
4338
4339               When used with "@", the repeat count represents an offset from
4340               the start of the innermost "()" group.
4341
4342               When used with ".", the repeat count determines the starting
4343               position to calculate the value offset as follows:
4344
4345               •   If the repeat count is 0, it's relative to the current
4346                   position.
4347
4348               •   If the repeat count is "*", the offset is relative to the
4349                   start of the packed string.
4350
4351               •   And if it's an integer n, the offset is relative to the
4352                   start of the nth innermost "( )" group, or to the start of
4353                   the string if n is bigger then the group level.
4354
4355               The repeat count for "u" is interpreted as the maximal number
4356               of bytes to encode per line of output, with 0, 1 and 2 replaced
4357               by 45.  The repeat count should not be more than 65.
4358
4359           •   The "a", "A", and "Z" types gobble just one value, but pack it
4360               as a string of length count, padding with nulls or spaces as
4361               needed.  When unpacking, "A" strips trailing whitespace and
4362               nulls, "Z" strips everything after the first null, and "a"
4363               returns data with no stripping at all.
4364
4365               If the value to pack is too long, the result is truncated.  If
4366               it's too long and an explicit count is provided, "Z" packs only
4367               "$count-1" bytes, followed by a null byte.  Thus "Z" always
4368               packs a trailing null, except when the count is 0.
4369
4370           •   Likewise, the "b" and "B" formats pack a string that's that
4371               many bits long.  Each such format generates 1 bit of the
4372               result.  These are typically followed by a repeat count like
4373               "B8" or "B64".
4374
4375               Each result bit is based on the least-significant bit of the
4376               corresponding input character, i.e., on "ord($char)%2".  In
4377               particular, characters "0" and "1" generate bits 0 and 1, as do
4378               characters "\000" and "\001".
4379
4380               Starting from the beginning of the input string, each 8-tuple
4381               of characters is converted to 1 character of output.  With
4382               format "b", the first character of the 8-tuple determines the
4383               least-significant bit of a character; with format "B", it
4384               determines the most-significant bit of a character.
4385
4386               If the length of the input string is not evenly divisible by 8,
4387               the remainder is packed as if the input string were padded by
4388               null characters at the end.  Similarly during unpacking,
4389               "extra" bits are ignored.
4390
4391               If the input string is longer than needed, remaining characters
4392               are ignored.
4393
4394               A "*" for the repeat count uses all characters of the input
4395               field.  On unpacking, bits are converted to a string of 0s and
4396               1s.
4397
4398           •   The "h" and "H" formats pack a string that many nybbles (4-bit
4399               groups, representable as hexadecimal digits, "0".."9" "a".."f")
4400               long.
4401
4402               For each such format, "pack" generates 4 bits of result.  With
4403               non-alphabetical characters, the result is based on the 4
4404               least-significant bits of the input character, i.e., on
4405               "ord($char)%16".  In particular, characters "0" and "1"
4406               generate nybbles 0 and 1, as do bytes "\000" and "\001".  For
4407               characters "a".."f" and "A".."F", the result is compatible with
4408               the usual hexadecimal digits, so that "a" and "A" both generate
4409               the nybble "0xA==10".  Use only these specific hex characters
4410               with this format.
4411
4412               Starting from the beginning of the template to "pack", each
4413               pair of characters is converted to 1 character of output.  With
4414               format "h", the first character of the pair determines the
4415               least-significant nybble of the output character; with format
4416               "H", it determines the most-significant nybble.
4417
4418               If the length of the input string is not even, it behaves as if
4419               padded by a null character at the end.  Similarly, "extra"
4420               nybbles are ignored during unpacking.
4421
4422               If the input string is longer than needed, extra characters are
4423               ignored.
4424
4425               A "*" for the repeat count uses all characters of the input
4426               field.  For "unpack", nybbles are converted to a string of
4427               hexadecimal digits.
4428
4429           •   The "p" format packs a pointer to a null-terminated string.
4430               You are responsible for ensuring that the string is not a
4431               temporary value, as that could potentially get deallocated
4432               before you got around to using the packed result.  The "P"
4433               format packs a pointer to a structure of the size indicated by
4434               the length.  A null pointer is created if the corresponding
4435               value for "p" or "P" is "undef"; similarly with "unpack", where
4436               a null pointer unpacks into "undef".
4437
4438               If your system has a strange pointer size--meaning a pointer is
4439               neither as big as an int nor as big as a long--it may not be
4440               possible to pack or unpack pointers in big- or little-endian
4441               byte order.  Attempting to do so raises an exception.
4442
4443           •   The "/" template character allows packing and unpacking of a
4444               sequence of items where the packed structure contains a packed
4445               item count followed by the packed items themselves.  This is
4446               useful when the structure you're unpacking has encoded the
4447               sizes or repeat counts for some of its fields within the
4448               structure itself as separate fields.
4449
4450               For "pack", you write length-item"/"sequence-item, and the
4451               length-item describes how the length value is packed.  Formats
4452               likely to be of most use are integer-packing ones like "n" for
4453               Java strings, "w" for ASN.1 or SNMP, and "N" for Sun XDR.
4454
4455               For "pack", sequence-item may have a repeat count, in which
4456               case the minimum of that and the number of available items is
4457               used as the argument for length-item.  If it has no repeat
4458               count or uses a '*', the number of available items is used.
4459
4460               For "unpack", an internal stack of integer arguments unpacked
4461               so far is used.  You write "/"sequence-item and the repeat
4462               count is obtained by popping off the last element from the
4463               stack.  The sequence-item must not have a repeat count.
4464
4465               If sequence-item refers to a string type ("A", "a", or "Z"),
4466               the length-item is the string length, not the number of
4467               strings.  With an explicit repeat count for pack, the packed
4468               string is adjusted to that length.  For example:
4469
4470                This code:                             gives this result:
4471
4472                unpack("W/a", "\004Gurusamy")          ("Guru")
4473                unpack("a3/A A*", "007 Bond  J ")      (" Bond", "J")
4474                unpack("a3 x2 /A A*", "007: Bond, J.") ("Bond, J", ".")
4475
4476                pack("n/a* w/a","hello,","world")     "\000\006hello,\005world"
4477                pack("a/W2", ord("a") .. ord("z"))    "2ab"
4478
4479               The length-item is not returned explicitly from "unpack".
4480
4481               Supplying a count to the length-item format letter is only
4482               useful with "A", "a", or "Z".  Packing with a length-item of
4483               "a" or "Z" may introduce "\000" characters, which Perl does not
4484               regard as legal in numeric strings.
4485
4486           •   The integer types "s", "S", "l", and "L" may be followed by a
4487               "!" modifier to specify native shorts or longs.  As shown in
4488               the example above, a bare "l" means exactly 32 bits, although
4489               the native "long" as seen by the local C compiler may be
4490               larger.  This is mainly an issue on 64-bit platforms.  You can
4491               see whether using "!" makes any difference this way:
4492
4493                   printf "format s is %d, s! is %d\n",
4494                       length pack("s"), length pack("s!");
4495
4496                   printf "format l is %d, l! is %d\n",
4497                       length pack("l"), length pack("l!");
4498
4499               "i!" and "I!" are also allowed, but only for completeness'
4500               sake: they are identical to "i" and "I".
4501
4502               The actual sizes (in bytes) of native shorts, ints, longs, and
4503               long longs on the platform where Perl was built are also
4504               available from the command line:
4505
4506                   $ perl -V:{short,int,long{,long}}size
4507                   shortsize='2';
4508                   intsize='4';
4509                   longsize='4';
4510                   longlongsize='8';
4511
4512               or programmatically via the "Config" module:
4513
4514                      use Config;
4515                      print $Config{shortsize},    "\n";
4516                      print $Config{intsize},      "\n";
4517                      print $Config{longsize},     "\n";
4518                      print $Config{longlongsize}, "\n";
4519
4520               $Config{longlongsize} is undefined on systems without long long
4521               support.
4522
4523           •   The integer formats "s", "S", "i", "I", "l", "L", "j", and "J"
4524               are inherently non-portable between processors and operating
4525               systems because they obey native byteorder and endianness.  For
4526               example, a 4-byte integer 0x12345678 (305419896 decimal) would
4527               be ordered natively (arranged in and handled by the CPU
4528               registers) into bytes as
4529
4530                   0x12 0x34 0x56 0x78  # big-endian
4531                   0x78 0x56 0x34 0x12  # little-endian
4532
4533               Basically, Intel and VAX CPUs are little-endian, while
4534               everybody else, including Motorola m68k/88k, PPC, Sparc, HP PA,
4535               Power, and Cray, are big-endian.  Alpha and MIPS can be either:
4536               Digital/Compaq uses (well, used) them in little-endian mode,
4537               but SGI/Cray uses them in big-endian mode.
4538
4539               The names big-endian and little-endian are comic references to
4540               the egg-eating habits of the little-endian Lilliputians and the
4541               big-endian Blefuscudians from the classic Jonathan Swift
4542               satire, Gulliver's Travels.  This entered computer lingo via
4543               the paper "On Holy Wars and a Plea for Peace" by Danny Cohen,
4544               USC/ISI IEN 137, April 1, 1980.
4545
4546               Some systems may have even weirder byte orders such as
4547
4548                  0x56 0x78 0x12 0x34
4549                  0x34 0x12 0x78 0x56
4550
4551               These are called mid-endian, middle-endian, mixed-endian, or
4552               just weird.
4553
4554               You can determine your system endianness with this incantation:
4555
4556                  printf("%#02x ", $_) for unpack("W*", pack L=>0x12345678);
4557
4558               The byteorder on the platform where Perl was built is also
4559               available via Config:
4560
4561                   use Config;
4562                   print "$Config{byteorder}\n";
4563
4564               or from the command line:
4565
4566                   $ perl -V:byteorder
4567
4568               Byteorders "1234" and "12345678" are little-endian; "4321" and
4569               "87654321" are big-endian.  Systems with multiarchitecture
4570               binaries will have "ffff", signifying that static information
4571               doesn't work, one must use runtime probing.
4572
4573               For portably packed integers, either use the formats "n", "N",
4574               "v", and "V" or else use the ">" and "<" modifiers described
4575               immediately below.  See also perlport.
4576
4577           •   Also floating point numbers have endianness.  Usually (but not
4578               always) this agrees with the integer endianness.  Even though
4579               most platforms these days use the IEEE 754 binary format, there
4580               are differences, especially if the long doubles are involved.
4581               You can see the "Config" variables "doublekind" and
4582               "longdblkind" (also "doublesize", "longdblsize"): the "kind"
4583               values are enums, unlike "byteorder".
4584
4585               Portability-wise the best option is probably to keep to the
4586               IEEE 754 64-bit doubles, and of agreed-upon endianness.
4587               Another possibility is the "%a") format of "printf".
4588
4589           •   Starting with Perl 5.10.0, integer and floating-point formats,
4590               along with the "p" and "P" formats and "()" groups, may all be
4591               followed by the ">" or "<" endianness modifiers to respectively
4592               enforce big- or little-endian byte-order.  These modifiers are
4593               especially useful given how "n", "N", "v", and "V" don't cover
4594               signed integers, 64-bit integers, or floating-point values.
4595
4596               Here are some concerns to keep in mind when using an endianness
4597               modifier:
4598
4599               •   Exchanging signed integers between different platforms
4600                   works only when all platforms store them in the same
4601                   format.  Most platforms store signed integers in two's-
4602                   complement notation, so usually this is not an issue.
4603
4604               •   The ">" or "<" modifiers can only be used on floating-point
4605                   formats on big- or little-endian machines.  Otherwise,
4606                   attempting to use them raises an exception.
4607
4608               •   Forcing big- or little-endian byte-order on floating-point
4609                   values for data exchange can work only if all platforms use
4610                   the same binary representation such as IEEE floating-point.
4611                   Even if all platforms are using IEEE, there may still be
4612                   subtle differences.  Being able to use ">" or "<" on
4613                   floating-point values can be useful, but also dangerous if
4614                   you don't know exactly what you're doing.  It is not a
4615                   general way to portably store floating-point values.
4616
4617               •   When using ">" or "<" on a "()" group, this affects all
4618                   types inside the group that accept byte-order modifiers,
4619                   including all subgroups.  It is silently ignored for all
4620                   other types.  You are not allowed to override the byte-
4621                   order within a group that already has a byte-order modifier
4622                   suffix.
4623
4624           •   Real numbers (floats and doubles) are in native machine format
4625               only.  Due to the multiplicity of floating-point formats and
4626               the lack of a standard "network" representation for them, no
4627               facility for interchange has been made.  This means that packed
4628               floating-point data written on one machine may not be readable
4629               on another, even if both use IEEE floating-point arithmetic
4630               (because the endianness of the memory representation is not
4631               part of the IEEE spec).  See also perlport.
4632
4633               If you know exactly what you're doing, you can use the ">" or
4634               "<" modifiers to force big- or little-endian byte-order on
4635               floating-point values.
4636
4637               Because Perl uses doubles (or long doubles, if configured)
4638               internally for all numeric calculation, converting from double
4639               into float and thence to double again loses precision, so
4640               "unpack("f", pack("f", $foo)") will not in general equal $foo.
4641
4642           •   Pack and unpack can operate in two modes: character mode ("C0"
4643               mode) where the packed string is processed per character, and
4644               UTF-8 byte mode ("U0" mode) where the packed string is
4645               processed in its UTF-8-encoded Unicode form on a byte-by-byte
4646               basis.  Character mode is the default unless the format string
4647               starts with "U".  You can always switch mode mid-format with an
4648               explicit "C0" or "U0" in the format.  This mode remains in
4649               effect until the next mode change, or until the end of the "()"
4650               group it (directly) applies to.
4651
4652               Using "C0" to get Unicode characters while using "U0" to get
4653               non-Unicode bytes is not necessarily obvious.   Probably only
4654               the first of these is what you want:
4655
4656                   $ perl -CS -E 'say "\x{3B1}\x{3C9}"' |
4657                     perl -CS -ne 'printf "%v04X\n", $_ for unpack("C0A*", $_)'
4658                   03B1.03C9
4659                   $ perl -CS -E 'say "\x{3B1}\x{3C9}"' |
4660                     perl -CS -ne 'printf "%v02X\n", $_ for unpack("U0A*", $_)'
4661                   CE.B1.CF.89
4662                   $ perl -CS -E 'say "\x{3B1}\x{3C9}"' |
4663                     perl -C0 -ne 'printf "%v02X\n", $_ for unpack("C0A*", $_)'
4664                   CE.B1.CF.89
4665                   $ perl -CS -E 'say "\x{3B1}\x{3C9}"' |
4666                     perl -C0 -ne 'printf "%v02X\n", $_ for unpack("U0A*", $_)'
4667                   C3.8E.C2.B1.C3.8F.C2.89
4668
4669               Those examples also illustrate that you should not try to use
4670               "pack"/"unpack" as a substitute for the Encode module.
4671
4672           •   You must yourself do any alignment or padding by inserting, for
4673               example, enough "x"es while packing.  There is no way for
4674               "pack" and "unpack" to know where characters are going to or
4675               coming from, so they handle their output and input as flat
4676               sequences of characters.
4677
4678           •   A "()" group is a sub-TEMPLATE enclosed in parentheses.  A
4679               group may take a repeat count either as postfix, or for
4680               "unpack", also via the "/" template character.  Within each
4681               repetition of a group, positioning with "@" starts over at 0.
4682               Therefore, the result of
4683
4684                   pack("@1A((@2A)@3A)", qw[X Y Z])
4685
4686               is the string "\0X\0\0YZ".
4687
4688           •   "x" and "X" accept the "!" modifier to act as alignment
4689               commands: they jump forward or back to the closest position
4690               aligned at a multiple of "count" characters.  For example, to
4691               "pack" or "unpack" a C structure like
4692
4693                   struct {
4694                       char   c;    /* one signed, 8-bit character */
4695                       double d;
4696                       char   cc[2];
4697                   }
4698
4699               one may need to use the template "c x![d] d c[2]".  This
4700               assumes that doubles must be aligned to the size of double.
4701
4702               For alignment commands, a "count" of 0 is equivalent to a
4703               "count" of 1; both are no-ops.
4704
4705           •   "n", "N", "v" and "V" accept the "!" modifier to represent
4706               signed 16-/32-bit integers in big-/little-endian order.  This
4707               is portable only when all platforms sharing packed data use the
4708               same binary representation for signed integers; for example,
4709               when all platforms use two's-complement representation.
4710
4711           •   Comments can be embedded in a TEMPLATE using "#" through the
4712               end of line.  White space can separate pack codes from each
4713               other, but modifiers and repeat counts must follow immediately.
4714               Breaking complex templates into individual line-by-line
4715               components, suitably annotated, can do as much to improve
4716               legibility and maintainability of pack/unpack formats as "/x"
4717               can for complicated pattern matches.
4718
4719           •   If TEMPLATE requires more arguments than "pack" is given,
4720               "pack" assumes additional "" arguments.  If TEMPLATE requires
4721               fewer arguments than given, extra arguments are ignored.
4722
4723           •   Attempting to pack the special floating point values "Inf" and
4724               "NaN" (infinity, also in negative, and not-a-number) into
4725               packed integer values (like "L") is a fatal error.  The reason
4726               for this is that there simply isn't any sensible mapping for
4727               these special values into integers.
4728
4729           Examples:
4730
4731               $foo = pack("WWWW",65,66,67,68);
4732               # foo eq "ABCD"
4733               $foo = pack("W4",65,66,67,68);
4734               # same thing
4735               $foo = pack("W4",0x24b6,0x24b7,0x24b8,0x24b9);
4736               # same thing with Unicode circled letters.
4737               $foo = pack("U4",0x24b6,0x24b7,0x24b8,0x24b9);
4738               # same thing with Unicode circled letters.  You don't get the
4739               # UTF-8 bytes because the U at the start of the format caused
4740               # a switch to U0-mode, so the UTF-8 bytes get joined into
4741               # characters
4742               $foo = pack("C0U4",0x24b6,0x24b7,0x24b8,0x24b9);
4743               # foo eq "\xe2\x92\xb6\xe2\x92\xb7\xe2\x92\xb8\xe2\x92\xb9"
4744               # This is the UTF-8 encoding of the string in the
4745               # previous example
4746
4747               $foo = pack("ccxxcc",65,66,67,68);
4748               # foo eq "AB\0\0CD"
4749
4750               # NOTE: The examples above featuring "W" and "c" are true
4751               # only on ASCII and ASCII-derived systems such as ISO Latin 1
4752               # and UTF-8.  On EBCDIC systems, the first example would be
4753               #      $foo = pack("WWWW",193,194,195,196);
4754
4755               $foo = pack("s2",1,2);
4756               # "\001\000\002\000" on little-endian
4757               # "\000\001\000\002" on big-endian
4758
4759               $foo = pack("a4","abcd","x","y","z");
4760               # "abcd"
4761
4762               $foo = pack("aaaa","abcd","x","y","z");
4763               # "axyz"
4764
4765               $foo = pack("a14","abcdefg");
4766               # "abcdefg\0\0\0\0\0\0\0"
4767
4768               $foo = pack("i9pl", gmtime);
4769               # a real struct tm (on my system anyway)
4770
4771               $utmp_template = "Z8 Z8 Z16 L";
4772               $utmp = pack($utmp_template, @utmp1);
4773               # a struct utmp (BSDish)
4774
4775               @utmp2 = unpack($utmp_template, $utmp);
4776               # "@utmp1" eq "@utmp2"
4777
4778               sub bintodec {
4779                   unpack("N", pack("B32", substr("0" x 32 . shift, -32)));
4780               }
4781
4782               $foo = pack('sx2l', 12, 34);
4783               # short 12, two zero bytes padding, long 34
4784               $bar = pack('s@4l', 12, 34);
4785               # short 12, zero fill to position 4, long 34
4786               # $foo eq $bar
4787               $baz = pack('s.l', 12, 4, 34);
4788               # short 12, zero fill to position 4, long 34
4789
4790               $foo = pack('nN', 42, 4711);
4791               # pack big-endian 16- and 32-bit unsigned integers
4792               $foo = pack('S>L>', 42, 4711);
4793               # exactly the same
4794               $foo = pack('s<l<', -42, 4711);
4795               # pack little-endian 16- and 32-bit signed integers
4796               $foo = pack('(sl)<', -42, 4711);
4797               # exactly the same
4798
4799           The same template may generally also be used in "unpack".
4800
4801       package NAMESPACE
4802       package NAMESPACE VERSION
4803       package NAMESPACE BLOCK
4804       package NAMESPACE VERSION BLOCK
4805           Declares the BLOCK or the rest of the compilation unit as being in
4806           the given namespace.  The scope of the package declaration is
4807           either the supplied code BLOCK or, in the absence of a BLOCK, from
4808           the declaration itself through the end of current scope (the
4809           enclosing block, file, or "eval").  That is, the forms without a
4810           BLOCK are operative through the end of the current scope, just like
4811           the "my", "state", and "our" operators.  All unqualified dynamic
4812           identifiers in this scope will be in the given namespace, except
4813           where overridden by another "package" declaration or when they're
4814           one of the special identifiers that qualify into "main::", like
4815           "STDOUT", "ARGV", "ENV", and the punctuation variables.
4816
4817           A package statement affects dynamic variables only, including those
4818           you've used "local" on, but not lexically-scoped variables, which
4819           are created with "my", "state", or "our".  Typically it would be
4820           the first declaration in a file included by "require" or "use".
4821           You can switch into a package in more than one place, since this
4822           only determines which default symbol table the compiler uses for
4823           the rest of that block.  You can refer to identifiers in other
4824           packages than the current one by prefixing the identifier with the
4825           package name and a double colon, as in $SomePack::var or
4826           "ThatPack::INPUT_HANDLE".  If package name is omitted, the "main"
4827           package is assumed.  That is, $::sail is equivalent to $main::sail
4828           (as well as to "$main'sail", still seen in ancient code, mostly
4829           from Perl 4).
4830
4831           If VERSION is provided, "package" sets the $VERSION variable in the
4832           given namespace to a version object with the VERSION provided.
4833           VERSION must be a "strict" style version number as defined by the
4834           version module: a positive decimal number (integer or decimal-
4835           fraction) without exponentiation or else a dotted-decimal v-string
4836           with a leading 'v' character and at least three components.  You
4837           should set $VERSION only once per package.
4838
4839           See "Packages" in perlmod for more information about packages,
4840           modules, and classes.  See perlsub for other scoping issues.
4841
4842       __PACKAGE__
4843           A special token that returns the name of the package in which it
4844           occurs.
4845
4846       pipe READHANDLE,WRITEHANDLE
4847           Opens a pair of connected pipes like the corresponding system call.
4848           Note that if you set up a loop of piped processes, deadlock can
4849           occur unless you are very careful.  In addition, note that Perl's
4850           pipes use IO buffering, so you may need to set $| to flush your
4851           WRITEHANDLE after each command, depending on the application.
4852
4853           Returns true on success.
4854
4855           See IPC::Open2, IPC::Open3, and "Bidirectional Communication with
4856           Another Process" in perlipc for examples of such things.
4857
4858           On systems that support a close-on-exec flag on files, that flag is
4859           set on all newly opened file descriptors whose "fileno"s are higher
4860           than the current value of $^F (by default 2 for "STDERR").  See
4861           "$^F" in perlvar.
4862
4863       pop ARRAY
4864       pop Pops and returns the last value of the array, shortening the array
4865           by one element.
4866
4867           Returns the undefined value if the array is empty, although this
4868           may also happen at other times.  If ARRAY is omitted, pops the
4869           @ARGV array in the main program, but the @_ array in subroutines,
4870           just like "shift".
4871
4872           Starting with Perl 5.14, an experimental feature allowed "pop" to
4873           take a scalar expression. This experiment has been deemed
4874           unsuccessful, and was removed as of Perl 5.24.
4875
4876       pos SCALAR
4877       pos Returns the offset of where the last "m//g" search left off for the
4878           variable in question ($_ is used when the variable is not
4879           specified).  This offset is in characters unless the (no-longer-
4880           recommended) "use bytes" pragma is in effect, in which case the
4881           offset is in bytes.  Note that 0 is a valid match offset.  "undef"
4882           indicates that the search position is reset (usually due to match
4883           failure, but can also be because no match has yet been run on the
4884           scalar).
4885
4886           "pos" directly accesses the location used by the regexp engine to
4887           store the offset, so assigning to "pos" will change that offset,
4888           and so will also influence the "\G" zero-width assertion in regular
4889           expressions.  Both of these effects take place for the next match,
4890           so you can't affect the position with "pos" during the current
4891           match, such as in "(?{pos() = 5})" or "s//pos() = 5/e".
4892
4893           Setting "pos" also resets the matched with zero-length flag,
4894           described under "Repeated Patterns Matching a Zero-length
4895           Substring" in perlre.
4896
4897           Because a failed "m//gc" match doesn't reset the offset, the return
4898           from "pos" won't change either in this case.  See perlre and
4899           perlop.
4900
4901       print FILEHANDLE LIST
4902       print FILEHANDLE
4903       print LIST
4904       print
4905           Prints a string or a list of strings.  Returns true if successful.
4906           FILEHANDLE may be a scalar variable containing the name of or a
4907           reference to the filehandle, thus introducing one level of
4908           indirection.  (NOTE: If FILEHANDLE is a variable and the next token
4909           is a term, it may be misinterpreted as an operator unless you
4910           interpose a "+" or put parentheses around the arguments.)  If
4911           FILEHANDLE is omitted, prints to the last selected (see "select")
4912           output handle.  If LIST is omitted, prints $_ to the currently
4913           selected output handle.  To use FILEHANDLE alone to print the
4914           content of $_ to it, you must use a bareword filehandle like "FH",
4915           not an indirect one like $fh.  To set the default output handle to
4916           something other than STDOUT, use the select operation.
4917
4918           The current value of $, (if any) is printed between each LIST item.
4919           The current value of "$\" (if any) is printed after the entire LIST
4920           has been printed.  Because print takes a LIST, anything in the LIST
4921           is evaluated in list context, including any subroutines whose
4922           return lists you pass to "print".  Be careful not to follow the
4923           print keyword with a left parenthesis unless you want the
4924           corresponding right parenthesis to terminate the arguments to the
4925           print; put parentheses around all arguments (or interpose a "+",
4926           but that doesn't look as good).
4927
4928           If you're storing handles in an array or hash, or in general
4929           whenever you're using any expression more complex than a bareword
4930           handle or a plain, unsubscripted scalar variable to retrieve it,
4931           you will have to use a block returning the filehandle value
4932           instead, in which case the LIST may not be omitted:
4933
4934               print { $files[$i] } "stuff\n";
4935               print { $OK ? *STDOUT : *STDERR } "stuff\n";
4936
4937           Printing to a closed pipe or socket will generate a SIGPIPE signal.
4938           See perlipc for more on signal handling.
4939
4940       printf FILEHANDLE FORMAT, LIST
4941       printf FILEHANDLE
4942       printf FORMAT, LIST
4943       printf
4944           Equivalent to "print FILEHANDLE sprintf(FORMAT, LIST)", except that
4945           "$\" (the output record separator) is not appended.  The FORMAT and
4946           the LIST are actually parsed as a single list.  The first argument
4947           of the list will be interpreted as the "printf" format.  This means
4948           that "printf(@_)" will use $_[0] as the format.  See sprintf for an
4949           explanation of the format argument.  If "use locale" (including
4950           "use locale ':not_characters'") is in effect and "POSIX::setlocale"
4951           has been called, the character used for the decimal separator in
4952           formatted floating-point numbers is affected by the "LC_NUMERIC"
4953           locale setting.  See perllocale and POSIX.
4954
4955           For historical reasons, if you omit the list, $_ is used as the
4956           format; to use FILEHANDLE without a list, you must use a bareword
4957           filehandle like "FH", not an indirect one like $fh.  However, this
4958           will rarely do what you want; if $_ contains formatting codes, they
4959           will be replaced with the empty string and a warning will be
4960           emitted if warnings are enabled.  Just use "print" if you want to
4961           print the contents of $_.
4962
4963           Don't fall into the trap of using a "printf" when a simple "print"
4964           would do.  The "print" is more efficient and less error prone.
4965
4966       prototype FUNCTION
4967       prototype
4968           Returns the prototype of a function as a string (or "undef" if the
4969           function has no prototype).  FUNCTION is a reference to, or the
4970           name of, the function whose prototype you want to retrieve.  If
4971           FUNCTION is omitted, $_ is used.
4972
4973           If FUNCTION is a string starting with "CORE::", the rest is taken
4974           as a name for a Perl builtin.  If the builtin's arguments cannot be
4975           adequately expressed by a prototype (such as "system"), "prototype"
4976           returns "undef", because the builtin does not really behave like a
4977           Perl function.  Otherwise, the string describing the equivalent
4978           prototype is returned.
4979
4980       push ARRAY,LIST
4981           Treats ARRAY as a stack by appending the values of LIST to the end
4982           of ARRAY.  The length of ARRAY increases by the length of LIST.
4983           Has the same effect as
4984
4985               for my $value (LIST) {
4986                   $ARRAY[++$#ARRAY] = $value;
4987               }
4988
4989           but is more efficient.  Returns the number of elements in the array
4990           following the completed "push".
4991
4992           Starting with Perl 5.14, an experimental feature allowed "push" to
4993           take a scalar expression. This experiment has been deemed
4994           unsuccessful, and was removed as of Perl 5.24.
4995
4996       q/STRING/
4997       qq/STRING/
4998       qw/STRING/
4999       qx/STRING/
5000           Generalized quotes.  See "Quote-Like Operators" in perlop.
5001
5002       qr/STRING/
5003           Regexp-like quote.  See "Regexp Quote-Like Operators" in perlop.
5004
5005       quotemeta EXPR
5006       quotemeta
5007           Returns the value of EXPR with all the ASCII non-"word" characters
5008           backslashed.  (That is, all ASCII characters not matching
5009           "/[A-Za-z_0-9]/" will be preceded by a backslash in the returned
5010           string, regardless of any locale settings.)  This is the internal
5011           function implementing the "\Q" escape in double-quoted strings.
5012           (See below for the behavior on non-ASCII code points.)
5013
5014           If EXPR is omitted, uses $_.
5015
5016           quotemeta (and "\Q" ... "\E") are useful when interpolating strings
5017           into regular expressions, because by default an interpolated
5018           variable will be considered a mini-regular expression.  For
5019           example:
5020
5021               my $sentence = 'The quick brown fox jumped over the lazy dog';
5022               my $substring = 'quick.*?fox';
5023               $sentence =~ s{$substring}{big bad wolf};
5024
5025           Will cause $sentence to become 'The big bad wolf jumped over...'.
5026
5027           On the other hand:
5028
5029               my $sentence = 'The quick brown fox jumped over the lazy dog';
5030               my $substring = 'quick.*?fox';
5031               $sentence =~ s{\Q$substring\E}{big bad wolf};
5032
5033           Or:
5034
5035               my $sentence = 'The quick brown fox jumped over the lazy dog';
5036               my $substring = 'quick.*?fox';
5037               my $quoted_substring = quotemeta($substring);
5038               $sentence =~ s{$quoted_substring}{big bad wolf};
5039
5040           Will both leave the sentence as is.  Normally, when accepting
5041           literal string input from the user, "quotemeta" or "\Q" must be
5042           used.
5043
5044           Beware that if you put literal backslashes (those not inside
5045           interpolated variables) between "\Q" and "\E", double-quotish
5046           backslash interpolation may lead to confusing results.  If you need
5047           to use literal backslashes within "\Q...\E", consult "Gory details
5048           of parsing quoted constructs" in perlop.
5049
5050           Because the result of "\Q STRING \E" has all metacharacters quoted,
5051           there is no way to insert a literal "$" or "@" inside a "\Q\E"
5052           pair.  If protected by "\", "$" will be quoted to become "\\\$"; if
5053           not, it is interpreted as the start of an interpolated scalar.
5054
5055           In Perl v5.14, all non-ASCII characters are quoted in
5056           non-UTF-8-encoded strings, but not quoted in UTF-8 strings.
5057
5058           Starting in Perl v5.16, Perl adopted a Unicode-defined strategy for
5059           quoting non-ASCII characters; the quoting of ASCII characters is
5060           unchanged.
5061
5062           Also unchanged is the quoting of non-UTF-8 strings when outside the
5063           scope of a "use feature 'unicode_strings'", which is to quote all
5064           characters in the upper Latin1 range.  This provides complete
5065           backwards compatibility for old programs which do not use Unicode.
5066           (Note that "unicode_strings" is automatically enabled within the
5067           scope of a "use v5.12" or greater.)
5068
5069           Within the scope of "use locale", all non-ASCII Latin1 code points
5070           are quoted whether the string is encoded as UTF-8 or not.  As
5071           mentioned above, locale does not affect the quoting of ASCII-range
5072           characters.  This protects against those locales where characters
5073           such as "|" are considered to be word characters.
5074
5075           Otherwise, Perl quotes non-ASCII characters using an adaptation
5076           from Unicode (see <https://www.unicode.org/reports/tr31/>).  The
5077           only code points that are quoted are those that have any of the
5078           Unicode properties:  Pattern_Syntax, Pattern_White_Space,
5079           White_Space, Default_Ignorable_Code_Point, or
5080           General_Category=Control.
5081
5082           Of these properties, the two important ones are Pattern_Syntax and
5083           Pattern_White_Space.  They have been set up by Unicode for exactly
5084           this purpose of deciding which characters in a regular expression
5085           pattern should be quoted.  No character that can be in an
5086           identifier has these properties.
5087
5088           Perl promises, that if we ever add regular expression pattern
5089           metacharacters to the dozen already defined ("\ | ( ) [ { ^ $ * + ?
5090           ."), that we will only use ones that have the Pattern_Syntax
5091           property.  Perl also promises, that if we ever add characters that
5092           are considered to be white space in regular expressions (currently
5093           mostly affected by "/x"), they will all have the
5094           Pattern_White_Space property.
5095
5096           Unicode promises that the set of code points that have these two
5097           properties will never change, so something that is not quoted in
5098           v5.16 will never need to be quoted in any future Perl release.
5099           (Not all the code points that match Pattern_Syntax have actually
5100           had characters assigned to them; so there is room to grow, but they
5101           are quoted whether assigned or not.  Perl, of course, would never
5102           use an unassigned code point as an actual metacharacter.)
5103
5104           Quoting characters that have the other 3 properties is done to
5105           enhance the readability of the regular expression and not because
5106           they actually need to be quoted for regular expression purposes
5107           (characters with the White_Space property are likely to be
5108           indistinguishable on the page or screen from those with the
5109           Pattern_White_Space property; and the other two properties contain
5110           non-printing characters).
5111
5112       rand EXPR
5113       rand
5114           Returns a random fractional number greater than or equal to 0 and
5115           less than the value of EXPR.  (EXPR should be positive.)  If EXPR
5116           is omitted, the value 1 is used.  Currently EXPR with the value 0
5117           is also special-cased as 1 (this was undocumented before Perl 5.8.0
5118           and is subject to change in future versions of Perl).
5119           Automatically calls "srand" unless "srand" has already been called.
5120           See also "srand".
5121
5122           Apply "int" to the value returned by "rand" if you want random
5123           integers instead of random fractional numbers.  For example,
5124
5125               int(rand(10))
5126
5127           returns a random integer between 0 and 9, inclusive.
5128
5129           (Note: If your rand function consistently returns numbers that are
5130           too large or too small, then your version of Perl was probably
5131           compiled with the wrong number of RANDBITS.)
5132
5133           "rand" is not cryptographically secure.  You should not rely on it
5134           in security-sensitive situations.  As of this writing, a number of
5135           third-party CPAN modules offer random number generators intended by
5136           their authors to be cryptographically secure, including:
5137           Data::Entropy, Crypt::Random, Math::Random::Secure, and
5138           Math::TrulyRandom.
5139
5140       read FILEHANDLE,SCALAR,LENGTH,OFFSET
5141       read FILEHANDLE,SCALAR,LENGTH
5142           Attempts to read LENGTH characters of data into variable SCALAR
5143           from the specified FILEHANDLE.  Returns the number of characters
5144           actually read, 0 at end of file, or undef if there was an error (in
5145           the latter case $! is also set).  SCALAR will be grown or shrunk so
5146           that the last character actually read is the last character of the
5147           scalar after the read.
5148
5149           An OFFSET may be specified to place the read data at some place in
5150           the string other than the beginning.  A negative OFFSET specifies
5151           placement at that many characters counting backwards from the end
5152           of the string.  A positive OFFSET greater than the length of SCALAR
5153           results in the string being padded to the required size with "\0"
5154           bytes before the result of the read is appended.
5155
5156           The call is implemented in terms of either Perl's or your system's
5157           native fread(3) library function, via the PerlIO layers applied to
5158           the handle.  To get a true read(2) system call, see sysread.
5159
5160           Note the characters: depending on the status of the filehandle,
5161           either (8-bit) bytes or characters are read.  By default, all
5162           filehandles operate on bytes, but for example if the filehandle has
5163           been opened with the ":utf8" I/O layer (see "open", and the open
5164           pragma), the I/O will operate on UTF8-encoded Unicode characters,
5165           not bytes.  Similarly for the ":encoding" layer: in that case
5166           pretty much any characters can be read.
5167
5168       readdir DIRHANDLE
5169           Returns the next directory entry for a directory opened by
5170           "opendir".  If used in list context, returns all the rest of the
5171           entries in the directory.  If there are no more entries, returns
5172           the undefined value in scalar context and the empty list in list
5173           context.
5174
5175           If you're planning to filetest the return values out of a
5176           "readdir", you'd better prepend the directory in question.
5177           Otherwise, because we didn't "chdir" there, it would have been
5178           testing the wrong file.
5179
5180               opendir(my $dh, $some_dir) || die "Can't opendir $some_dir: $!";
5181               my @dots = grep { /^\./ && -f "$some_dir/$_" } readdir($dh);
5182               closedir $dh;
5183
5184           As of Perl 5.12 you can use a bare "readdir" in a "while" loop,
5185           which will set $_ on every iteration.  If either a "readdir"
5186           expression or an explicit assignment of a "readdir" expression to a
5187           scalar is used as a "while"/"for" condition, then the condition
5188           actually tests for definedness of the expression's value, not for
5189           its regular truth value.
5190
5191               opendir(my $dh, $some_dir) || die "Can't open $some_dir: $!";
5192               while (readdir $dh) {
5193                   print "$some_dir/$_\n";
5194               }
5195               closedir $dh;
5196
5197           To avoid confusing would-be users of your code who are running
5198           earlier versions of Perl with mysterious failures, put this sort of
5199           thing at the top of your file to signal that your code will work
5200           only on Perls of a recent vintage:
5201
5202               use v5.12; # so readdir assigns to $_ in a lone while test
5203
5204       readline EXPR
5205       readline
5206           Reads from the filehandle whose typeglob is contained in EXPR (or
5207           from *ARGV if EXPR is not provided).  In scalar context, each call
5208           reads and returns the next line until end-of-file is reached,
5209           whereupon the subsequent call returns "undef".  In list context,
5210           reads until end-of-file is reached and returns a list of lines.
5211           Note that the notion of "line" used here is whatever you may have
5212           defined with $/ (or $INPUT_RECORD_SEPARATOR in English).  See "$/"
5213           in perlvar.
5214
5215           When $/ is set to "undef", when "readline" is in scalar context
5216           (i.e., file slurp mode), and when an empty file is read, it returns
5217           '' the first time, followed by "undef" subsequently.
5218
5219           This is the internal function implementing the "<EXPR>" operator,
5220           but you can use it directly.  The "<EXPR>" operator is discussed in
5221           more detail in "I/O Operators" in perlop.
5222
5223               my $line = <STDIN>;
5224               my $line = readline(STDIN);    # same thing
5225
5226           If "readline" encounters an operating system error, $! will be set
5227           with the corresponding error message.  It can be helpful to check
5228           $! when you are reading from filehandles you don't trust, such as a
5229           tty or a socket.  The following example uses the operator form of
5230           "readline" and dies if the result is not defined.
5231
5232               while ( ! eof($fh) ) {
5233                   defined( $_ = readline $fh ) or die "readline failed: $!";
5234                   ...
5235               }
5236
5237           Note that you can't handle "readline" errors that way with the
5238           "ARGV" filehandle.  In that case, you have to open each element of
5239           @ARGV yourself since "eof" handles "ARGV" differently.
5240
5241               foreach my $arg (@ARGV) {
5242                   open(my $fh, $arg) or warn "Can't open $arg: $!";
5243
5244                   while ( ! eof($fh) ) {
5245                       defined( $_ = readline $fh )
5246                           or die "readline failed for $arg: $!";
5247                       ...
5248                   }
5249               }
5250
5251           Like the "<EXPR>" operator, if a "readline" expression is used as
5252           the condition of a "while" or "for" loop, then it will be
5253           implicitly assigned to $_.  If either a "readline" expression or an
5254           explicit assignment of a "readline" expression to a scalar is used
5255           as a "while"/"for" condition, then the condition actually tests for
5256           definedness of the expression's value, not for its regular truth
5257           value.
5258
5259       readlink EXPR
5260       readlink
5261           Returns the value of a symbolic link, if symbolic links are
5262           implemented.  If not, raises an exception.  If there is a system
5263           error, returns the undefined value and sets $! (errno).  If EXPR is
5264           omitted, uses $_.
5265
5266           Portability issues: "readlink" in perlport.
5267
5268       readpipe EXPR
5269       readpipe
5270           EXPR is executed as a system command.  The collected standard
5271           output of the command is returned.  In scalar context, it comes
5272           back as a single (potentially multi-line) string.  In list context,
5273           returns a list of lines (however you've defined lines with $/ (or
5274           $INPUT_RECORD_SEPARATOR in English)).  This is the internal
5275           function implementing the "qx/EXPR/" operator, but you can use it
5276           directly.  The "qx/EXPR/" operator is discussed in more detail in
5277           ""qx/STRING/"" in perlop.  If EXPR is omitted, uses $_.
5278
5279       recv SOCKET,SCALAR,LENGTH,FLAGS
5280           Receives a message on a socket.  Attempts to receive LENGTH
5281           characters of data into variable SCALAR from the specified SOCKET
5282           filehandle.  SCALAR will be grown or shrunk to the length actually
5283           read.  Takes the same flags as the system call of the same name.
5284           Returns the address of the sender if SOCKET's protocol supports
5285           this; returns an empty string otherwise.  If there's an error,
5286           returns the undefined value.  This call is actually implemented in
5287           terms of the recvfrom(2) system call.  See "UDP: Message Passing"
5288           in perlipc for examples.
5289
5290           Note that if the socket has been marked as ":utf8", "recv" will
5291           throw an exception.  The ":encoding(...)" layer implicitly
5292           introduces the ":utf8" layer.  See "binmode".
5293
5294       redo LABEL
5295       redo EXPR
5296       redo
5297           The "redo" command restarts the loop block without evaluating the
5298           conditional again.  The "continue" block, if any, is not executed.
5299           If the LABEL is omitted, the command refers to the innermost
5300           enclosing loop.  The "redo EXPR" form, available starting in Perl
5301           5.18.0, allows a label name to be computed at run time, and is
5302           otherwise identical to "redo LABEL".  Programs that want to lie to
5303           themselves about what was just input normally use this command:
5304
5305               # a simpleminded Pascal comment stripper
5306               # (warning: assumes no { or } in strings)
5307               LINE: while (<STDIN>) {
5308                   while (s|({.*}.*){.*}|$1 |) {}
5309                   s|{.*}| |;
5310                   if (s|{.*| |) {
5311                       my $front = $_;
5312                       while (<STDIN>) {
5313                           if (/}/) {  # end of comment?
5314                               s|^|$front\{|;
5315                               redo LINE;
5316                           }
5317                       }
5318                   }
5319                   print;
5320               }
5321
5322           "redo" cannot return a value from a block that typically returns a
5323           value, such as "eval {}", "sub {}", or "do {}". It will perform its
5324           flow control behavior, which precludes any return value. It should
5325           not be used to exit a "grep" or "map" operation.
5326
5327           Note that a block by itself is semantically identical to a loop
5328           that executes once.  Thus "redo" inside such a block will
5329           effectively turn it into a looping construct.
5330
5331           See also "continue" for an illustration of how "last", "next", and
5332           "redo" work.
5333
5334           Unlike most named operators, this has the same precedence as
5335           assignment.  It is also exempt from the looks-like-a-function rule,
5336           so "redo ("foo")."bar"" will cause "bar" to be part of the argument
5337           to "redo".
5338
5339       ref EXPR
5340       ref Examines the value of EXPR, expecting it to be a reference, and
5341           returns a string giving information about the reference and the
5342           type of referent.  If EXPR is not specified, $_ will be used.
5343
5344           If the operand is not a reference, then the empty string will be
5345           returned.  An empty string will only be returned in this situation.
5346           "ref" is often useful to just test whether a value is a reference,
5347           which can be done by comparing the result to the empty string.  It
5348           is a common mistake to use the result of "ref" directly as a truth
5349           value: this goes wrong because 0 (which is false) can be returned
5350           for a reference.
5351
5352           If the operand is a reference to a blessed object, then the name of
5353           the class into which the referent is blessed will be returned.
5354           "ref" doesn't care what the physical type of the referent is;
5355           blessing takes precedence over such concerns.  Beware that exact
5356           comparison of "ref" results against a class name doesn't perform a
5357           class membership test: a class's members also include objects
5358           blessed into subclasses, for which "ref" will return the name of
5359           the subclass.  Also beware that class names can clash with the
5360           built-in type names (described below).
5361
5362           If the operand is a reference to an unblessed object, then the
5363           return value indicates the type of object.  If the unblessed
5364           referent is not a scalar, then the return value will be one of the
5365           strings "ARRAY", "HASH", "CODE", "FORMAT", or "IO", indicating only
5366           which kind of object it is.  If the unblessed referent is a scalar,
5367           then the return value will be one of the strings "SCALAR",
5368           "VSTRING", "REF", "GLOB", "LVALUE", or "REGEXP", depending on the
5369           kind of value the scalar currently has.   But note that "qr//"
5370           scalars are created already blessed, so "ref qr/.../" will likely
5371           return "Regexp".  Beware that these built-in type names can also be
5372           used as class names, so "ref" returning one of these names doesn't
5373           unambiguously indicate that the referent is of the kind to which
5374           the name refers.
5375
5376           The ambiguity between built-in type names and class names
5377           significantly limits the utility of "ref".  For unambiguous
5378           information, use "Scalar::Util::blessed()" for information about
5379           blessing, and "Scalar::Util::reftype()" for information about
5380           physical types.  Use the "isa" method for class membership tests,
5381           though one must be sure of blessedness before attempting a method
5382           call.  Alternatively, the "isa" operator can test class membership
5383           without checking blessedness first.
5384
5385           See also perlref and perlobj.
5386
5387       rename OLDNAME,NEWNAME
5388           Changes the name of a file; an existing file NEWNAME will be
5389           clobbered.  Returns true for success; on failure returns false and
5390           sets $!.
5391
5392           Behavior of this function varies wildly depending on your system
5393           implementation.  For example, it will usually not work across file
5394           system boundaries, even though the system mv command sometimes
5395           compensates for this.  Other restrictions include whether it works
5396           on directories, open files, or pre-existing files.  Check perlport
5397           and either the rename(2) manpage or equivalent system documentation
5398           for details.
5399
5400           For a platform independent "move" function look at the File::Copy
5401           module.
5402
5403           Portability issues: "rename" in perlport.
5404
5405       require VERSION
5406       require EXPR
5407       require
5408           Demands a version of Perl specified by VERSION, or demands some
5409           semantics specified by EXPR or by $_ if EXPR is not supplied.
5410
5411           VERSION may be either a literal such as v5.24.1, which will be
5412           compared to $^V (or $PERL_VERSION in English), or a numeric
5413           argument of the form 5.024001, which will be compared to $]. An
5414           exception is raised if VERSION is greater than the version of the
5415           current Perl interpreter.  Compare with "use", which can do a
5416           similar check at compile time.
5417
5418           Specifying VERSION as a numeric argument of the form 5.024001
5419           should generally be avoided as older less readable syntax compared
5420           to v5.24.1. Before perl 5.8.0 (released in 2002), the more verbose
5421           numeric form was the only supported syntax, which is why you might
5422           see it in older code.
5423
5424               require v5.24.1;    # run time version check
5425               require 5.24.1;     # ditto
5426               require 5.024_001;  # ditto; older syntax compatible
5427                                     with perl 5.6
5428
5429           Otherwise, "require" demands that a library file be included if it
5430           hasn't already been included.  The file is included via the do-FILE
5431           mechanism, which is essentially just a variety of "eval" with the
5432           caveat that lexical variables in the invoking script will be
5433           invisible to the included code.  If it were implemented in pure
5434           Perl, it would have semantics similar to the following:
5435
5436               use Carp 'croak';
5437               use version;
5438
5439               sub require {
5440                   my ($filename) = @_;
5441                   if ( my $version = eval { version->parse($filename) } ) {
5442                       if ( $version > $^V ) {
5443                          my $vn = $version->normal;
5444                          croak "Perl $vn required--this is only $^V, stopped";
5445                       }
5446                       return 1;
5447                   }
5448
5449                   if (exists $INC{$filename}) {
5450                       return 1 if $INC{$filename};
5451                       croak "Compilation failed in require";
5452                   }
5453
5454                   foreach $prefix (@INC) {
5455                       if (ref($prefix)) {
5456                           #... do other stuff - see text below ....
5457                       }
5458                       # (see text below about possible appending of .pmc
5459                       # suffix to $filename)
5460                       my $realfilename = "$prefix/$filename";
5461                       next if ! -e $realfilename || -d _ || -b _;
5462                       $INC{$filename} = $realfilename;
5463                       my $result = do($realfilename);
5464                                    # but run in caller's namespace
5465
5466                       if (!defined $result) {
5467                           $INC{$filename} = undef;
5468                           croak $@ ? "$@Compilation failed in require"
5469                                    : "Can't locate $filename: $!\n";
5470                       }
5471                       if (!$result) {
5472                           delete $INC{$filename};
5473                           croak "$filename did not return true value";
5474                       }
5475                       $! = 0;
5476                       return $result;
5477                   }
5478                   croak "Can't locate $filename in \@INC ...";
5479               }
5480
5481           Note that the file will not be included twice under the same
5482           specified name.
5483
5484           The file must return true as the last statement to indicate
5485           successful execution of any initialization code, so it's customary
5486           to end such a file with "1;" unless you're sure it'll return true
5487           otherwise.  But it's better just to put the "1;", in case you add
5488           more statements.
5489
5490           If EXPR is a bareword, "require" assumes a .pm extension and
5491           replaces "::" with "/" in the filename for you, to make it easy to
5492           load standard modules.  This form of loading of modules does not
5493           risk altering your namespace, however it will autovivify the stash
5494           for the required module.
5495
5496           In other words, if you try this:
5497
5498                   require Foo::Bar;     # a splendid bareword
5499
5500           The require function will actually look for the Foo/Bar.pm file in
5501           the directories specified in the @INC array, and it will autovivify
5502           the "Foo::Bar::" stash at compile time.
5503
5504           But if you try this:
5505
5506                   my $class = 'Foo::Bar';
5507                   require $class;       # $class is not a bareword
5508               #or
5509                   require "Foo::Bar";   # not a bareword because of the ""
5510
5511           The require function will look for the Foo::Bar file in the @INC
5512           array and will complain about not finding Foo::Bar there.  In this
5513           case you can do:
5514
5515                   eval "require $class";
5516
5517           or you could do
5518
5519                   require "Foo/Bar.pm";
5520
5521           Neither of these forms will autovivify any stashes at compile time
5522           and only have run time effects.
5523
5524           Now that you understand how "require" looks for files with a
5525           bareword argument, there is a little extra functionality going on
5526           behind the scenes.  Before "require" looks for a .pm extension, it
5527           will first look for a similar filename with a .pmc extension.  If
5528           this file is found, it will be loaded in place of any file ending
5529           in a .pm extension. This applies to both the explicit "require
5530           "Foo/Bar.pm";" form and the "require Foo::Bar;" form.
5531
5532           You can also insert hooks into the import facility by putting Perl
5533           code directly into the @INC array.  There are three forms of hooks:
5534           subroutine references, array references, and blessed objects.
5535
5536           Subroutine references are the simplest case.  When the inclusion
5537           system walks through @INC and encounters a subroutine, this
5538           subroutine gets called with two parameters, the first a reference
5539           to itself, and the second the name of the file to be included
5540           (e.g., Foo/Bar.pm).  The subroutine should return either nothing or
5541           else a list of up to four values in the following order:
5542
5543           1.  A reference to a scalar, containing any initial source code to
5544               prepend to the file or generator output.
5545
5546           2.  A filehandle, from which the file will be read.
5547
5548           3.  A reference to a subroutine.  If there is no filehandle
5549               (previous item), then this subroutine is expected to generate
5550               one line of source code per call, writing the line into $_ and
5551               returning 1, then finally at end of file returning 0.  If there
5552               is a filehandle, then the subroutine will be called to act as a
5553               simple source filter, with the line as read in $_.  Again,
5554               return 1 for each valid line, and 0 after all lines have been
5555               returned.  For historical reasons the subroutine will receive a
5556               meaningless argument (in fact always the numeric value zero) as
5557               $_[0].
5558
5559           4.  Optional state for the subroutine.  The state is passed in as
5560               $_[1].
5561
5562           If an empty list, "undef", or nothing that matches the first 3
5563           values above is returned, then "require" looks at the remaining
5564           elements of @INC.  Note that this filehandle must be a real
5565           filehandle (strictly a typeglob or reference to a typeglob, whether
5566           blessed or unblessed); tied filehandles will be ignored and
5567           processing will stop there.
5568
5569           If the hook is an array reference, its first element must be a
5570           subroutine reference.  This subroutine is called as above, but the
5571           first parameter is the array reference.  This lets you indirectly
5572           pass arguments to the subroutine.
5573
5574           In other words, you can write:
5575
5576               push @INC, \&my_sub;
5577               sub my_sub {
5578                   my ($coderef, $filename) = @_;  # $coderef is \&my_sub
5579                   ...
5580               }
5581
5582           or:
5583
5584               push @INC, [ \&my_sub, $x, $y, ... ];
5585               sub my_sub {
5586                   my ($arrayref, $filename) = @_;
5587                   # Retrieve $x, $y, ...
5588                   my (undef, @parameters) = @$arrayref;
5589                   ...
5590               }
5591
5592           If the hook is an object, it must provide an "INC" method that will
5593           be called as above, the first parameter being the object itself.
5594           (Note that you must fully qualify the sub's name, as unqualified
5595           "INC" is always forced into package "main".)  Here is a typical
5596           code layout:
5597
5598               # In Foo.pm
5599               package Foo;
5600               sub new { ... }
5601               sub Foo::INC {
5602                   my ($self, $filename) = @_;
5603                   ...
5604               }
5605
5606               # In the main program
5607               push @INC, Foo->new(...);
5608
5609           These hooks are also permitted to set the %INC entry corresponding
5610           to the files they have loaded.  See "%INC" in perlvar.
5611
5612           For a yet-more-powerful import facility, see "use" and perlmod.
5613
5614       reset EXPR
5615       reset
5616           Generally used in a "continue" block at the end of a loop to clear
5617           variables and reset "m?pattern?" searches so that they work again.
5618           The expression is interpreted as a list of single characters
5619           (hyphens allowed for ranges).  All variables (scalars, arrays, and
5620           hashes) in the current package beginning with one of those letters
5621           are reset to their pristine state.  If the expression is omitted,
5622           one-match searches ("m?pattern?") are reset to match again.  Only
5623           resets variables or searches in the current package.  Always
5624           returns 1.  Examples:
5625
5626               reset 'X';      # reset all X variables
5627               reset 'a-z';    # reset lower case variables
5628               reset;          # just reset m?one-time? searches
5629
5630           Resetting "A-Z" is not recommended because you'll wipe out your
5631           @ARGV and @INC arrays and your %ENV hash.
5632
5633           Resets only package variables; lexical variables are unaffected,
5634           but they clean themselves up on scope exit anyway, so you'll
5635           probably want to use them instead.  See "my".
5636
5637       return EXPR
5638       return
5639           Returns from a subroutine, "eval", "do FILE", "sort" block or regex
5640           eval block (but not a "grep", "map", or "do BLOCK" block) with the
5641           value given in EXPR.  Evaluation of EXPR may be in list, scalar, or
5642           void context, depending on how the return value will be used, and
5643           the context may vary from one execution to the next (see
5644           "wantarray").  If no EXPR is given, returns an empty list in list
5645           context, the undefined value in scalar context, and (of course)
5646           nothing at all in void context.
5647
5648           (In the absence of an explicit "return", a subroutine, "eval", or
5649           "do FILE" automatically returns the value of the last expression
5650           evaluated.)
5651
5652           Unlike most named operators, this is also exempt from the looks-
5653           like-a-function rule, so "return ("foo")."bar"" will cause "bar" to
5654           be part of the argument to "return".
5655
5656       reverse LIST
5657           In list context, returns a list value consisting of the elements of
5658           LIST in the opposite order.  In scalar context, concatenates the
5659           elements of LIST and returns a string value with all characters in
5660           the opposite order.
5661
5662               print join(", ", reverse "world", "Hello"); # Hello, world
5663
5664               print scalar reverse "dlrow ,", "olleH";    # Hello, world
5665
5666           Used without arguments in scalar context, "reverse" reverses $_.
5667
5668               $_ = "dlrow ,olleH";
5669               print reverse;                         # No output, list context
5670               print scalar reverse;                  # Hello, world
5671
5672           Note that reversing an array to itself (as in "@a = reverse @a")
5673           will preserve non-existent elements whenever possible; i.e., for
5674           non-magical arrays or for tied arrays with "EXISTS" and "DELETE"
5675           methods.
5676
5677           This operator is also handy for inverting a hash, although there
5678           are some caveats.  If a value is duplicated in the original hash,
5679           only one of those can be represented as a key in the inverted hash.
5680           Also, this has to unwind one hash and build a whole new one, which
5681           may take some time on a large hash, such as from a DBM file.
5682
5683               my %by_name = reverse %by_address;  # Invert the hash
5684
5685       rewinddir DIRHANDLE
5686           Sets the current position to the beginning of the directory for the
5687           "readdir" routine on DIRHANDLE.
5688
5689           Portability issues: "rewinddir" in perlport.
5690
5691       rindex STR,SUBSTR,POSITION
5692       rindex STR,SUBSTR
5693           Works just like "index" except that it returns the position of the
5694           last occurrence of SUBSTR in STR.  If POSITION is specified,
5695           returns the last occurrence beginning at or before that position.
5696
5697       rmdir FILENAME
5698       rmdir
5699           Deletes the directory specified by FILENAME if that directory is
5700           empty.  If it succeeds it returns true; otherwise it returns false
5701           and sets $! (errno).  If FILENAME is omitted, uses $_.
5702
5703           To remove a directory tree recursively ("rm -rf" on Unix) look at
5704           the "rmtree" function of the File::Path module.
5705
5706       s///
5707           The substitution operator.  See "Regexp Quote-Like Operators" in
5708           perlop.
5709
5710       say FILEHANDLE LIST
5711       say FILEHANDLE
5712       say LIST
5713       say Just like "print", but implicitly appends a newline at the end of
5714           the LIST instead of any value "$\" might have.  To use FILEHANDLE
5715           without a LIST to print the contents of $_ to it, you must use a
5716           bareword filehandle like "FH", not an indirect one like $fh.
5717
5718           "say" is available only if the "say" feature is enabled or if it is
5719           prefixed with "CORE::".  The "say" feature is enabled automatically
5720           with a "use v5.10" (or higher) declaration in the current scope.
5721
5722       scalar EXPR
5723           Forces EXPR to be interpreted in scalar context and returns the
5724           value of EXPR.
5725
5726               my @counts = ( scalar @a, scalar @b, scalar @c );
5727
5728           There is no equivalent operator to force an expression to be
5729           interpolated in list context because in practice, this is never
5730           needed.  If you really wanted to do so, however, you could use the
5731           construction "@{[ (some expression) ]}", but usually a simple
5732           "(some expression)" suffices.
5733
5734           Because "scalar" is a unary operator, if you accidentally use a
5735           parenthesized list for the EXPR, this behaves as a scalar comma
5736           expression, evaluating all but the last element in void context and
5737           returning the final element evaluated in scalar context.  This is
5738           seldom what you want.
5739
5740           The following single statement:
5741
5742               print uc(scalar(foo(), $bar)), $baz;
5743
5744           is the moral equivalent of these two:
5745
5746               foo();
5747               print(uc($bar), $baz);
5748
5749           See perlop for more details on unary operators and the comma
5750           operator, and perldata for details on evaluating a hash in scalar
5751           context.
5752
5753       seek FILEHANDLE,POSITION,WHENCE
5754           Sets FILEHANDLE's position, just like the fseek(3) call of C
5755           "stdio".  FILEHANDLE may be an expression whose value gives the
5756           name of the filehandle.  The values for WHENCE are 0 to set the new
5757           position in bytes to POSITION; 1 to set it to the current position
5758           plus POSITION; and 2 to set it to EOF plus POSITION, typically
5759           negative.  For WHENCE you may use the constants "SEEK_SET",
5760           "SEEK_CUR", and "SEEK_END" (start of the file, current position,
5761           end of the file) from the Fcntl module.  Returns 1 on success,
5762           false otherwise.
5763
5764           Note the emphasis on bytes: even if the filehandle has been set to
5765           operate on characters (for example using the ":encoding(UTF-8)" I/O
5766           layer), the "seek", "tell", and "sysseek" family of functions use
5767           byte offsets, not character offsets, because seeking to a character
5768           offset would be very slow in a UTF-8 file.
5769
5770           If you want to position the file for "sysread" or "syswrite", don't
5771           use "seek", because buffering makes its effect on the file's read-
5772           write position unpredictable and non-portable.  Use "sysseek"
5773           instead.
5774
5775           Due to the rules and rigors of ANSI C, on some systems you have to
5776           do a seek whenever you switch between reading and writing.  Amongst
5777           other things, this may have the effect of calling stdio's
5778           clearerr(3).  A WHENCE of 1 ("SEEK_CUR") is useful for not moving
5779           the file position:
5780
5781               seek($fh, 0, 1);
5782
5783           This is also useful for applications emulating "tail -f".  Once you
5784           hit EOF on your read and then sleep for a while, you (probably)
5785           have to stick in a dummy "seek" to reset things.  The "seek"
5786           doesn't change the position, but it does clear the end-of-file
5787           condition on the handle, so that the next "readline FILE" makes
5788           Perl try again to read something.  (We hope.)
5789
5790           If that doesn't work (some I/O implementations are particularly
5791           cantankerous), you might need something like this:
5792
5793               for (;;) {
5794                   for ($curpos = tell($fh); $_ = readline($fh);
5795                        $curpos = tell($fh)) {
5796                       # search for some stuff and put it into files
5797                   }
5798                   sleep($for_a_while);
5799                   seek($fh, $curpos, 0);
5800               }
5801
5802       seekdir DIRHANDLE,POS
5803           Sets the current position for the "readdir" routine on DIRHANDLE.
5804           POS must be a value returned by "telldir".  "seekdir" also has the
5805           same caveats about possible directory compaction as the
5806           corresponding system library routine.
5807
5808       select FILEHANDLE
5809       select
5810           Returns the currently selected filehandle.  If FILEHANDLE is
5811           supplied, sets the new current default filehandle for output.  This
5812           has two effects: first, a "write" or a "print" without a filehandle
5813           default to this FILEHANDLE.  Second, references to variables
5814           related to output will refer to this output channel.
5815
5816           For example, to set the top-of-form format for more than one output
5817           channel, you might do the following:
5818
5819               select(REPORT1);
5820               $^ = 'report1_top';
5821               select(REPORT2);
5822               $^ = 'report2_top';
5823
5824           FILEHANDLE may be an expression whose value gives the name of the
5825           actual filehandle.  Thus:
5826
5827               my $oldfh = select(STDERR); $| = 1; select($oldfh);
5828
5829           Some programmers may prefer to think of filehandles as objects with
5830           methods, preferring to write the last example as:
5831
5832               STDERR->autoflush(1);
5833
5834           (Prior to Perl version 5.14, you have to "use IO::Handle;"
5835           explicitly first.)
5836
5837           Portability issues: "select" in perlport.
5838
5839       select RBITS,WBITS,EBITS,TIMEOUT
5840           This calls the select(2) syscall with the bit masks specified,
5841           which can be constructed using "fileno" and "vec", along these
5842           lines:
5843
5844               my $rin = my $win = my $ein = '';
5845               vec($rin, fileno(STDIN),  1) = 1;
5846               vec($win, fileno(STDOUT), 1) = 1;
5847               $ein = $rin | $win;
5848
5849           If you want to select on many filehandles, you may wish to write a
5850           subroutine like this:
5851
5852               sub fhbits {
5853                   my @fhlist = @_;
5854                   my $bits = "";
5855                   for my $fh (@fhlist) {
5856                       vec($bits, fileno($fh), 1) = 1;
5857                   }
5858                   return $bits;
5859               }
5860               my $rin = fhbits(\*STDIN, $tty, $mysock);
5861
5862           The usual idiom is:
5863
5864            my ($nfound, $timeleft) =
5865              select(my $rout = $rin, my $wout = $win, my $eout = $ein,
5866                                                                     $timeout);
5867
5868           or to block until something becomes ready just do this
5869
5870            my $nfound =
5871              select(my $rout = $rin, my $wout = $win, my $eout = $ein, undef);
5872
5873           Most systems do not bother to return anything useful in $timeleft,
5874           so calling "select" in scalar context just returns $nfound.
5875
5876           Any of the bit masks can also be "undef".  The timeout, if
5877           specified, is in seconds, which may be fractional.  Note: not all
5878           implementations are capable of returning the $timeleft.  If not,
5879           they always return $timeleft equal to the supplied $timeout.
5880
5881           You can effect a sleep of 250 milliseconds this way:
5882
5883               select(undef, undef, undef, 0.25);
5884
5885           Note that whether "select" gets restarted after signals (say,
5886           SIGALRM) is implementation-dependent.  See also perlport for notes
5887           on the portability of "select".
5888
5889           On error, "select" behaves just like select(2): it returns "-1" and
5890           sets $!.
5891
5892           On some Unixes, select(2) may report a socket file descriptor as
5893           "ready for reading" even when no data is available, and thus any
5894           subsequent "read" would block.  This can be avoided if you always
5895           use "O_NONBLOCK" on the socket.  See select(2) and fcntl(2) for
5896           further details.
5897
5898           The standard "IO::Select" module provides a user-friendlier
5899           interface to "select", mostly because it does all the bit-mask work
5900           for you.
5901
5902           WARNING: One should not attempt to mix buffered I/O (like "read" or
5903           "readline") with "select", except as permitted by POSIX, and even
5904           then only on POSIX systems.  You have to use "sysread" instead.
5905
5906           Portability issues: "select" in perlport.
5907
5908       semctl ID,SEMNUM,CMD,ARG
5909           Calls the System V IPC function semctl(2).  You'll probably have to
5910           say
5911
5912               use IPC::SysV;
5913
5914           first to get the correct constant definitions.  If CMD is IPC_STAT
5915           or GETALL, then ARG must be a variable that will hold the returned
5916           semid_ds structure or semaphore value array.  Returns like "ioctl":
5917           the undefined value for error, ""0 but true"" for zero, or the
5918           actual return value otherwise.  The ARG must consist of a vector of
5919           native short integers, which may be created with
5920           "pack("s!",(0)x$nsem)".  See also "SysV IPC" in perlipc and the
5921           documentation for "IPC::SysV" and "IPC::Semaphore".
5922
5923           Portability issues: "semctl" in perlport.
5924
5925       semget KEY,NSEMS,FLAGS
5926           Calls the System V IPC function semget(2).  Returns the semaphore
5927           id, or the undefined value on error.  See also "SysV IPC" in
5928           perlipc and the documentation for "IPC::SysV" and "IPC::Semaphore".
5929
5930           Portability issues: "semget" in perlport.
5931
5932       semop KEY,OPSTRING
5933           Calls the System V IPC function semop(2) for semaphore operations
5934           such as signalling and waiting.  OPSTRING must be a packed array of
5935           semop structures.  Each semop structure can be generated with
5936           "pack("s!3", $semnum, $semop, $semflag)".  The length of OPSTRING
5937           implies the number of semaphore operations.  Returns true if
5938           successful, false on error.  As an example, the following code
5939           waits on semaphore $semnum of semaphore id $semid:
5940
5941               my $semop = pack("s!3", $semnum, -1, 0);
5942               die "Semaphore trouble: $!\n" unless semop($semid, $semop);
5943
5944           To signal the semaphore, replace "-1" with 1.  See also "SysV IPC"
5945           in perlipc and the documentation for "IPC::SysV" and
5946           "IPC::Semaphore".
5947
5948           Portability issues: "semop" in perlport.
5949
5950       send SOCKET,MSG,FLAGS,TO
5951       send SOCKET,MSG,FLAGS
5952           Sends a message on a socket.  Attempts to send the scalar MSG to
5953           the SOCKET filehandle.  Takes the same flags as the system call of
5954           the same name.  On unconnected sockets, you must specify a
5955           destination to send to, in which case it does a sendto(2) syscall.
5956           Returns the number of characters sent, or the undefined value on
5957           error.  The sendmsg(2) syscall is currently unimplemented.  See
5958           "UDP: Message Passing" in perlipc for examples.
5959
5960           Note that if the socket has been marked as ":utf8", "send" will
5961           throw an exception.  The ":encoding(...)" layer implicitly
5962           introduces the ":utf8" layer.  See "binmode".
5963
5964       setpgrp PID,PGRP
5965           Sets the current process group for the specified PID, 0 for the
5966           current process.  Raises an exception when used on a machine that
5967           doesn't implement POSIX setpgid(2) or BSD setpgrp(2).  If the
5968           arguments are omitted, it defaults to "0,0".  Note that the BSD 4.2
5969           version of "setpgrp" does not accept any arguments, so only
5970           "setpgrp(0,0)" is portable.  See also "POSIX::setsid()".
5971
5972           Portability issues: "setpgrp" in perlport.
5973
5974       setpriority WHICH,WHO,PRIORITY
5975           Sets the current priority for a process, a process group, or a
5976           user.  (See setpriority(2).)  Raises an exception when used on a
5977           machine that doesn't implement setpriority(2).
5978
5979           "WHICH" can be any of "PRIO_PROCESS", "PRIO_PGRP" or "PRIO_USER"
5980           imported from "RESOURCE CONSTANTS" in POSIX.
5981
5982           Portability issues: "setpriority" in perlport.
5983
5984       setsockopt SOCKET,LEVEL,OPTNAME,OPTVAL
5985           Sets the socket option requested.  Returns "undef" on error.  Use
5986           integer constants provided by the "Socket" module for LEVEL and
5987           OPNAME.  Values for LEVEL can also be obtained from getprotobyname.
5988           OPTVAL might either be a packed string or an integer.  An integer
5989           OPTVAL is shorthand for pack("i", OPTVAL).
5990
5991           An example disabling Nagle's algorithm on a socket:
5992
5993               use Socket qw(IPPROTO_TCP TCP_NODELAY);
5994               setsockopt($socket, IPPROTO_TCP, TCP_NODELAY, 1);
5995
5996           Portability issues: "setsockopt" in perlport.
5997
5998       shift ARRAY
5999       shift
6000           Shifts the first value of the array off and returns it, shortening
6001           the array by 1 and moving everything down.  If there are no
6002           elements in the array, returns the undefined value.  If ARRAY is
6003           omitted, shifts the @_ array within the lexical scope of
6004           subroutines and formats, and the @ARGV array outside a subroutine
6005           and also within the lexical scopes established by the "eval
6006           STRING", "BEGIN {}", "INIT {}", "CHECK {}", "UNITCHECK {}", and
6007           "END {}" constructs.
6008
6009           Starting with Perl 5.14, an experimental feature allowed "shift" to
6010           take a scalar expression. This experiment has been deemed
6011           unsuccessful, and was removed as of Perl 5.24.
6012
6013           See also "unshift", "push", and "pop".  "shift" and "unshift" do
6014           the same thing to the left end of an array that "pop" and "push" do
6015           to the right end.
6016
6017       shmctl ID,CMD,ARG
6018           Calls the System V IPC function shmctl.  You'll probably have to
6019           say
6020
6021               use IPC::SysV;
6022
6023           first to get the correct constant definitions.  If CMD is
6024           "IPC_STAT", then ARG must be a variable that will hold the returned
6025           "shmid_ds" structure.  Returns like ioctl: "undef" for error; "0
6026           but true" for zero; and the actual return value otherwise.  See
6027           also "SysV IPC" in perlipc and the documentation for "IPC::SysV".
6028
6029           Portability issues: "shmctl" in perlport.
6030
6031       shmget KEY,SIZE,FLAGS
6032           Calls the System V IPC function shmget.  Returns the shared memory
6033           segment id, or "undef" on error.  See also "SysV IPC" in perlipc
6034           and the documentation for "IPC::SysV".
6035
6036           Portability issues: "shmget" in perlport.
6037
6038       shmread ID,VAR,POS,SIZE
6039       shmwrite ID,STRING,POS,SIZE
6040           Reads or writes the System V shared memory segment ID starting at
6041           position POS for size SIZE by attaching to it, copying in/out, and
6042           detaching from it.  When reading, VAR must be a variable that will
6043           hold the data read.  When writing, if STRING is too long, only SIZE
6044           bytes are used; if STRING is too short, nulls are written to fill
6045           out SIZE bytes.  Return true if successful, false on error.
6046           "shmread" taints the variable.  See also "SysV IPC" in perlipc and
6047           the documentation for "IPC::SysV" and the "IPC::Shareable" module
6048           from CPAN.
6049
6050           Portability issues: "shmread" in perlport and "shmwrite" in
6051           perlport.
6052
6053       shutdown SOCKET,HOW
6054           Shuts down a socket connection in the manner indicated by HOW,
6055           which has the same interpretation as in the syscall of the same
6056           name.
6057
6058               shutdown($socket, 0);    # I/we have stopped reading data
6059               shutdown($socket, 1);    # I/we have stopped writing data
6060               shutdown($socket, 2);    # I/we have stopped using this socket
6061
6062           This is useful with sockets when you want to tell the other side
6063           you're done writing but not done reading, or vice versa.  It's also
6064           a more insistent form of close because it also disables the file
6065           descriptor in any forked copies in other processes.
6066
6067           Returns 1 for success; on error, returns "undef" if the first
6068           argument is not a valid filehandle, or returns 0 and sets $! for
6069           any other failure.
6070
6071       sin EXPR
6072       sin Returns the sine of EXPR (expressed in radians).  If EXPR is
6073           omitted, returns sine of $_.
6074
6075           For the inverse sine operation, you may use the "Math::Trig::asin"
6076           function, or use this relation:
6077
6078               sub asin { atan2($_[0], sqrt(1 - $_[0] * $_[0])) }
6079
6080       sleep EXPR
6081       sleep
6082           Causes the script to sleep for (integer) EXPR seconds, or forever
6083           if no argument is given.  Returns the integer number of seconds
6084           actually slept.
6085
6086           EXPR should be a positive integer. If called with a negative
6087           integer, "sleep" does not sleep but instead emits a warning, sets
6088           $! ("errno"), and returns zero.
6089
6090           "sleep 0" is permitted, but a function call to the underlying
6091           platform implementation still occurs, with any side effects that
6092           may have.  "sleep 0" is therefore not exactly identical to not
6093           sleeping at all.
6094
6095           May be interrupted if the process receives a signal such as
6096           "SIGALRM".
6097
6098               eval {
6099                   local $SIG{ALRM} = sub { die "Alarm!\n" };
6100                   sleep;
6101               };
6102               die $@ unless $@ eq "Alarm!\n";
6103
6104           You probably cannot mix "alarm" and "sleep" calls, because "sleep"
6105           is often implemented using "alarm".
6106
6107           On some older systems, it may sleep up to a full second less than
6108           what you requested, depending on how it counts seconds.  Most
6109           modern systems always sleep the full amount.  They may appear to
6110           sleep longer than that, however, because your process might not be
6111           scheduled right away in a busy multitasking system.
6112
6113           For delays of finer granularity than one second, the Time::HiRes
6114           module (from CPAN, and starting from Perl 5.8 part of the standard
6115           distribution) provides "usleep".  You may also use Perl's four-
6116           argument version of "select" leaving the first three arguments
6117           undefined, or you might be able to use the "syscall" interface to
6118           access setitimer(2) if your system supports it.  See perlfaq8 for
6119           details.
6120
6121           See also the POSIX module's "pause" function.
6122
6123       socket SOCKET,DOMAIN,TYPE,PROTOCOL
6124           Opens a socket of the specified kind and attaches it to filehandle
6125           SOCKET.  DOMAIN, TYPE, and PROTOCOL are specified the same as for
6126           the syscall of the same name.  You should "use Socket" first to get
6127           the proper definitions imported.  See the examples in "Sockets:
6128           Client/Server Communication" in perlipc.
6129
6130           On systems that support a close-on-exec flag on files, the flag
6131           will be set for the newly opened file descriptor, as determined by
6132           the value of $^F.  See "$^F" in perlvar.
6133
6134       socketpair SOCKET1,SOCKET2,DOMAIN,TYPE,PROTOCOL
6135           Creates an unnamed pair of sockets in the specified domain, of the
6136           specified type.  DOMAIN, TYPE, and PROTOCOL are specified the same
6137           as for the syscall of the same name.  If unimplemented, raises an
6138           exception.  Returns true if successful.
6139
6140           On systems that support a close-on-exec flag on files, the flag
6141           will be set for the newly opened file descriptors, as determined by
6142           the value of $^F.  See "$^F" in perlvar.
6143
6144           Some systems define "pipe" in terms of "socketpair", in which a
6145           call to "pipe($rdr, $wtr)" is essentially:
6146
6147               use Socket;
6148               socketpair(my $rdr, my $wtr, AF_UNIX, SOCK_STREAM, PF_UNSPEC);
6149               shutdown($rdr, 1);        # no more writing for reader
6150               shutdown($wtr, 0);        # no more reading for writer
6151
6152           See perlipc for an example of socketpair use.  Perl 5.8 and later
6153           will emulate socketpair using IP sockets to localhost if your
6154           system implements sockets but not socketpair.
6155
6156           Portability issues: "socketpair" in perlport.
6157
6158       sort SUBNAME LIST
6159       sort BLOCK LIST
6160       sort LIST
6161           In list context, this sorts the LIST and returns the sorted list
6162           value.  In scalar context, the behaviour of "sort" is undefined.
6163
6164           If SUBNAME or BLOCK is omitted, "sort"s in standard string
6165           comparison order.  If SUBNAME is specified, it gives the name of a
6166           subroutine that returns an integer less than, equal to, or greater
6167           than 0, depending on how the elements of the list are to be
6168           ordered.  (The "<=>" and "cmp" operators are extremely useful in
6169           such routines.)  SUBNAME may be a scalar variable name
6170           (unsubscripted), in which case the value provides the name of (or a
6171           reference to) the actual subroutine to use.  In place of a SUBNAME,
6172           you can provide a BLOCK as an anonymous, in-line sort subroutine.
6173
6174           If the subroutine's prototype is "($$)", the elements to be
6175           compared are passed by reference in @_, as for a normal subroutine.
6176           This is slower than unprototyped subroutines, where the elements to
6177           be compared are passed into the subroutine as the package global
6178           variables $a and $b (see example below).
6179
6180           If the subroutine is an XSUB, the elements to be compared are
6181           pushed on to the stack, the way arguments are usually passed to
6182           XSUBs.  $a and $b are not set.
6183
6184           The values to be compared are always passed by reference and should
6185           not be modified.
6186
6187           You also cannot exit out of the sort block or subroutine using any
6188           of the loop control operators described in perlsyn or with "goto".
6189
6190           When "use locale" (but not "use locale ':not_characters'") is in
6191           effect, "sort LIST" sorts LIST according to the current collation
6192           locale.  See perllocale.
6193
6194           "sort" returns aliases into the original list, much as a for loop's
6195           index variable aliases the list elements.  That is, modifying an
6196           element of a list returned by "sort" (for example, in a "foreach",
6197           "map" or "grep") actually modifies the element in the original
6198           list.  This is usually something to be avoided when writing clear
6199           code.
6200
6201           Historically Perl has varied in whether sorting is stable by
6202           default.  If stability matters, it can be controlled explicitly by
6203           using the sort pragma.
6204
6205           Examples:
6206
6207               # sort lexically
6208               my @articles = sort @files;
6209
6210               # same thing, but with explicit sort routine
6211               my @articles = sort {$a cmp $b} @files;
6212
6213               # now case-insensitively
6214               my @articles = sort {fc($a) cmp fc($b)} @files;
6215
6216               # same thing in reversed order
6217               my @articles = sort {$b cmp $a} @files;
6218
6219               # sort numerically ascending
6220               my @articles = sort {$a <=> $b} @files;
6221
6222               # sort numerically descending
6223               my @articles = sort {$b <=> $a} @files;
6224
6225               # this sorts the %age hash by value instead of key
6226               # using an in-line function
6227               my @eldest = sort { $age{$b} <=> $age{$a} } keys %age;
6228
6229               # sort using explicit subroutine name
6230               sub byage {
6231                   $age{$a} <=> $age{$b};  # presuming numeric
6232               }
6233               my @sortedclass = sort byage @class;
6234
6235               sub backwards { $b cmp $a }
6236               my @harry  = qw(dog cat x Cain Abel);
6237               my @george = qw(gone chased yz Punished Axed);
6238               print sort @harry;
6239                   # prints AbelCaincatdogx
6240               print sort backwards @harry;
6241                   # prints xdogcatCainAbel
6242               print sort @george, 'to', @harry;
6243                   # prints AbelAxedCainPunishedcatchaseddoggonetoxyz
6244
6245               # inefficiently sort by descending numeric compare using
6246               # the first integer after the first = sign, or the
6247               # whole record case-insensitively otherwise
6248
6249               my @new = sort {
6250                   ($b =~ /=(\d+)/)[0] <=> ($a =~ /=(\d+)/)[0]
6251                                       ||
6252                               fc($a)  cmp  fc($b)
6253               } @old;
6254
6255               # same thing, but much more efficiently;
6256               # we'll build auxiliary indices instead
6257               # for speed
6258               my (@nums, @caps);
6259               for (@old) {
6260                   push @nums, ( /=(\d+)/ ? $1 : undef );
6261                   push @caps, fc($_);
6262               }
6263
6264               my @new = @old[ sort {
6265                                      $nums[$b] <=> $nums[$a]
6266                                               ||
6267                                      $caps[$a] cmp $caps[$b]
6268                                    } 0..$#old
6269                             ];
6270
6271               # same thing, but without any temps
6272               my @new = map { $_->[0] }
6273                      sort { $b->[1] <=> $a->[1]
6274                                      ||
6275                             $a->[2] cmp $b->[2]
6276                      } map { [$_, /=(\d+)/, fc($_)] } @old;
6277
6278               # using a prototype allows you to use any comparison subroutine
6279               # as a sort subroutine (including other package's subroutines)
6280               package Other;
6281               sub backwards ($$) { $_[1] cmp $_[0]; }  # $a and $b are
6282                                                        # not set here
6283               package main;
6284               my @new = sort Other::backwards @old;
6285
6286               ## using a prototype with function signature
6287               use feature 'signatures';
6288               sub function_with_signature :prototype($$) ($one, $two) {
6289                   return $one <=> $two
6290               }
6291
6292               my @new = sort function_with_signature @old;
6293
6294               # guarantee stability
6295               use sort 'stable';
6296               my @new = sort { substr($a, 3, 5) cmp substr($b, 3, 5) } @old;
6297
6298           Warning: syntactical care is required when sorting the list
6299           returned from a function.  If you want to sort the list returned by
6300           the function call "find_records(@key)", you can use:
6301
6302               my @contact = sort { $a cmp $b } find_records @key;
6303               my @contact = sort +find_records(@key);
6304               my @contact = sort &find_records(@key);
6305               my @contact = sort(find_records(@key));
6306
6307           If instead you want to sort the array @key with the comparison
6308           routine "find_records()" then you can use:
6309
6310               my @contact = sort { find_records() } @key;
6311               my @contact = sort find_records(@key);
6312               my @contact = sort(find_records @key);
6313               my @contact = sort(find_records (@key));
6314
6315           $a and $b are set as package globals in the package the sort() is
6316           called from.  That means $main::a and $main::b (or $::a and $::b)
6317           in the "main" package, $FooPack::a and $FooPack::b in the "FooPack"
6318           package, etc.  If the sort block is in scope of a "my" or "state"
6319           declaration of $a and/or $b, you must spell out the full name of
6320           the variables in the sort block :
6321
6322              package main;
6323              my $a = "C"; # DANGER, Will Robinson, DANGER !!!
6324
6325              print sort { $a cmp $b }               qw(A C E G B D F H);
6326                                                     # WRONG
6327              sub badlexi { $a cmp $b }
6328              print sort badlexi                     qw(A C E G B D F H);
6329                                                     # WRONG
6330              # the above prints BACFEDGH or some other incorrect ordering
6331
6332              print sort { $::a cmp $::b }           qw(A C E G B D F H);
6333                                                     # OK
6334              print sort { our $a cmp our $b }       qw(A C E G B D F H);
6335                                                     # also OK
6336              print sort { our ($a, $b); $a cmp $b } qw(A C E G B D F H);
6337                                                     # also OK
6338              sub lexi { our $a cmp our $b }
6339              print sort lexi                        qw(A C E G B D F H);
6340                                                     # also OK
6341              # the above print ABCDEFGH
6342
6343           With proper care you may mix package and my (or state) $a and/or
6344           $b:
6345
6346              my $a = {
6347                 tiny   => -2,
6348                 small  => -1,
6349                 normal => 0,
6350                 big    => 1,
6351                 huge   => 2
6352              };
6353
6354              say sort { $a->{our $a} <=> $a->{our $b} }
6355                  qw{ huge normal tiny small big};
6356
6357              # prints tinysmallnormalbighuge
6358
6359           $a and $b are implicitly local to the sort() execution and regain
6360           their former values upon completing the sort.
6361
6362           Sort subroutines written using $a and $b are bound to their calling
6363           package. It is possible, but of limited interest, to define them in
6364           a different package, since the subroutine must still refer to the
6365           calling package's $a and $b :
6366
6367              package Foo;
6368              sub lexi { $Bar::a cmp $Bar::b }
6369              package Bar;
6370              ... sort Foo::lexi ...
6371
6372           Use the prototyped versions (see above) for a more generic
6373           alternative.
6374
6375           The comparison function is required to behave.  If it returns
6376           inconsistent results (sometimes saying $x[1] is less than $x[2] and
6377           sometimes saying the opposite, for example) the results are not
6378           well-defined.
6379
6380           Because "<=>" returns "undef" when either operand is "NaN" (not-a-
6381           number), be careful when sorting with a comparison function like
6382           "$a <=> $b" any lists that might contain a "NaN".  The following
6383           example takes advantage that "NaN != NaN" to eliminate any "NaN"s
6384           from the input list.
6385
6386               my @result = sort { $a <=> $b } grep { $_ == $_ } @input;
6387
6388           In this version of perl, the "sort" function is implemented via the
6389           mergesort algorithm.
6390
6391       splice ARRAY,OFFSET,LENGTH,LIST
6392       splice ARRAY,OFFSET,LENGTH
6393       splice ARRAY,OFFSET
6394       splice ARRAY
6395           Removes the elements designated by OFFSET and LENGTH from an array,
6396           and replaces them with the elements of LIST, if any.  In list
6397           context, returns the elements removed from the array.  In scalar
6398           context, returns the last element removed, or "undef" if no
6399           elements are removed.  The array grows or shrinks as necessary.  If
6400           OFFSET is negative then it starts that far from the end of the
6401           array.  If LENGTH is omitted, removes everything from OFFSET
6402           onward.  If LENGTH is negative, removes the elements from OFFSET
6403           onward except for -LENGTH elements at the end of the array.  If
6404           both OFFSET and LENGTH are omitted, removes everything.  If OFFSET
6405           is past the end of the array and a LENGTH was provided, Perl issues
6406           a warning, and splices at the end of the array.
6407
6408           The following equivalences hold (assuming "$#a >= $i" )
6409
6410               push(@a,$x,$y)      splice(@a,@a,0,$x,$y)
6411               pop(@a)             splice(@a,-1)
6412               shift(@a)           splice(@a,0,1)
6413               unshift(@a,$x,$y)   splice(@a,0,0,$x,$y)
6414               $a[$i] = $y         splice(@a,$i,1,$y)
6415
6416           "splice" can be used, for example, to implement n-ary queue
6417           processing:
6418
6419               sub nary_print {
6420                 my $n = shift;
6421                 while (my @next_n = splice @_, 0, $n) {
6422                   say join q{ -- }, @next_n;
6423                 }
6424               }
6425
6426               nary_print(3, qw(a b c d e f g h));
6427               # prints:
6428               #   a -- b -- c
6429               #   d -- e -- f
6430               #   g -- h
6431
6432           Starting with Perl 5.14, an experimental feature allowed "splice"
6433           to take a scalar expression. This experiment has been deemed
6434           unsuccessful, and was removed as of Perl 5.24.
6435
6436       split /PATTERN/,EXPR,LIMIT
6437       split /PATTERN/,EXPR
6438       split /PATTERN/
6439       split
6440           Splits the string EXPR into a list of strings and returns the list
6441           in list context, or the size of the list in scalar context.  (Prior
6442           to Perl 5.11, it also overwrote @_ with the list in void and scalar
6443           context. If you target old perls, beware.)
6444
6445           If only PATTERN is given, EXPR defaults to $_.
6446
6447           Anything in EXPR that matches PATTERN is taken to be a separator
6448           that separates the EXPR into substrings (called "fields") that do
6449           not include the separator.  Note that a separator may be longer
6450           than one character or even have no characters at all (the empty
6451           string, which is a zero-width match).
6452
6453           The PATTERN need not be constant; an expression may be used to
6454           specify a pattern that varies at runtime.
6455
6456           If PATTERN matches the empty string, the EXPR is split at the match
6457           position (between characters).  As an example, the following:
6458
6459               my @x = split(/b/, "abc"); # ("a", "c")
6460
6461           uses the "b" in 'abc' as a separator to produce the list ("a",
6462           "c").  However, this:
6463
6464               my @x = split(//, "abc"); # ("a", "b", "c")
6465
6466           uses empty string matches as separators; thus, the empty string may
6467           be used to split EXPR into a list of its component characters.
6468
6469           As a special case for "split", the empty pattern given in match
6470           operator syntax ("//") specifically matches the empty string, which
6471           is contrary to its usual interpretation as the last successful
6472           match.
6473
6474           If PATTERN is "/^/", then it is treated as if it used the multiline
6475           modifier ("/^/m"), since it isn't much use otherwise.
6476
6477           "/m" and any of the other pattern modifiers valid for "qr"
6478           (summarized in "qr/STRING/msixpodualn" in perlop) may be specified
6479           explicitly.
6480
6481           As another special case, "split" emulates the default behavior of
6482           the command line tool awk when the PATTERN is either omitted or a
6483           string composed of a single space character (such as ' ' or "\x20",
6484           but not e.g. "/ /").  In this case, any leading whitespace in EXPR
6485           is removed before splitting occurs, and the PATTERN is instead
6486           treated as if it were "/\s+/"; in particular, this means that any
6487           contiguous whitespace (not just a single space character) is used
6488           as a separator.
6489
6490               my @x = split(" ", "  Quick brown fox\n");
6491               # ("Quick", "brown", "fox")
6492
6493               my @x = split(" ", "RED\tGREEN\tBLUE");
6494               # ("RED", "GREEN", "BLUE")
6495
6496           Using split in this fashion is very similar to how "qw//" works.
6497
6498           However, this special treatment can be avoided by specifying the
6499           pattern "/ /" instead of the string " ", thereby allowing only a
6500           single space character to be a separator.  In earlier Perls this
6501           special case was restricted to the use of a plain " " as the
6502           pattern argument to split; in Perl 5.18.0 and later this special
6503           case is triggered by any expression which evaluates to the simple
6504           string " ".
6505
6506           As of Perl 5.28, this special-cased whitespace splitting works as
6507           expected in the scope of "use feature 'unicode_strings'". In
6508           previous versions, and outside the scope of that feature, it
6509           exhibits "The "Unicode Bug"" in perlunicode: characters that are
6510           whitespace according to Unicode rules but not according to ASCII
6511           rules can be treated as part of fields rather than as field
6512           separators, depending on the string's internal encoding.
6513
6514           If omitted, PATTERN defaults to a single space, " ", triggering the
6515           previously described awk emulation.
6516
6517           If LIMIT is specified and positive, it represents the maximum
6518           number of fields into which the EXPR may be split; in other words,
6519           LIMIT is one greater than the maximum number of times EXPR may be
6520           split.  Thus, the LIMIT value 1 means that EXPR may be split a
6521           maximum of zero times, producing a maximum of one field (namely,
6522           the entire value of EXPR).  For instance:
6523
6524               my @x = split(//, "abc", 1); # ("abc")
6525               my @x = split(//, "abc", 2); # ("a", "bc")
6526               my @x = split(//, "abc", 3); # ("a", "b", "c")
6527               my @x = split(//, "abc", 4); # ("a", "b", "c")
6528
6529           If LIMIT is negative, it is treated as if it were instead
6530           arbitrarily large; as many fields as possible are produced.
6531
6532           If LIMIT is omitted (or, equivalently, zero), then it is usually
6533           treated as if it were instead negative but with the exception that
6534           trailing empty fields are stripped (empty leading fields are always
6535           preserved); if all fields are empty, then all fields are considered
6536           to be trailing (and are thus stripped in this case).  Thus, the
6537           following:
6538
6539               my @x = split(/,/, "a,b,c,,,"); # ("a", "b", "c")
6540
6541           produces only a three element list.
6542
6543               my @x = split(/,/, "a,b,c,,,", -1); # ("a", "b", "c", "", "", "")
6544
6545           produces a six element list.
6546
6547           In time-critical applications, it is worthwhile to avoid splitting
6548           into more fields than necessary.  Thus, when assigning to a list,
6549           if LIMIT is omitted (or zero), then LIMIT is treated as though it
6550           were one larger than the number of variables in the list; for the
6551           following, LIMIT is implicitly 3:
6552
6553               my ($login, $passwd) = split(/:/);
6554
6555           Note that splitting an EXPR that evaluates to the empty string
6556           always produces zero fields, regardless of the LIMIT specified.
6557
6558           An empty leading field is produced when there is a positive-width
6559           match at the beginning of EXPR.  For instance:
6560
6561               my @x = split(/ /, " abc"); # ("", "abc")
6562
6563           splits into two elements.  However, a zero-width match at the
6564           beginning of EXPR never produces an empty field, so that:
6565
6566               my @x = split(//, " abc"); # (" ", "a", "b", "c")
6567
6568           splits into four elements instead of five.
6569
6570           An empty trailing field, on the other hand, is produced when there
6571           is a match at the end of EXPR, regardless of the length of the
6572           match (of course, unless a non-zero LIMIT is given explicitly, such
6573           fields are removed, as in the last example).  Thus:
6574
6575               my @x = split(//, " abc", -1); # (" ", "a", "b", "c", "")
6576
6577           If the PATTERN contains capturing groups, then for each separator,
6578           an additional field is produced for each substring captured by a
6579           group (in the order in which the groups are specified, as per
6580           backreferences); if any group does not match, then it captures the
6581           "undef" value instead of a substring.  Also, note that any such
6582           additional field is produced whenever there is a separator (that
6583           is, whenever a split occurs), and such an additional field does not
6584           count towards the LIMIT.  Consider the following expressions
6585           evaluated in list context (each returned list is provided in the
6586           associated comment):
6587
6588               my @x = split(/-|,/    , "1-10,20", 3);
6589               # ("1", "10", "20")
6590
6591               my @x = split(/(-|,)/  , "1-10,20", 3);
6592               # ("1", "-", "10", ",", "20")
6593
6594               my @x = split(/-|(,)/  , "1-10,20", 3);
6595               # ("1", undef, "10", ",", "20")
6596
6597               my @x = split(/(-)|,/  , "1-10,20", 3);
6598               # ("1", "-", "10", undef, "20")
6599
6600               my @x = split(/(-)|(,)/, "1-10,20", 3);
6601               # ("1", "-", undef, "10", undef, ",", "20")
6602
6603       sprintf FORMAT, LIST
6604           Returns a string formatted by the usual "printf" conventions of the
6605           C library function "sprintf".  See below for more details and see
6606           sprintf(3) or printf(3) on your system for an explanation of the
6607           general principles.
6608
6609           For example:
6610
6611                   # Format number with up to 8 leading zeroes
6612                   my $result = sprintf("%08d", $number);
6613
6614                   # Round number to 3 digits after decimal point
6615                   my $rounded = sprintf("%.3f", $number);
6616
6617           Perl does its own "sprintf" formatting: it emulates the C function
6618           sprintf(3), but doesn't use it except for floating-point numbers,
6619           and even then only standard modifiers are allowed.  Non-standard
6620           extensions in your local sprintf(3) are therefore unavailable from
6621           Perl.
6622
6623           Unlike "printf", "sprintf" does not do what you probably mean when
6624           you pass it an array as your first argument.  The array is given
6625           scalar context, and instead of using the 0th element of the array
6626           as the format, Perl will use the count of elements in the array as
6627           the format, which is almost never useful.
6628
6629           Perl's "sprintf" permits the following universally-known
6630           conversions:
6631
6632              %%    a percent sign
6633              %c    a character with the given number
6634              %s    a string
6635              %d    a signed integer, in decimal
6636              %u    an unsigned integer, in decimal
6637              %o    an unsigned integer, in octal
6638              %x    an unsigned integer, in hexadecimal
6639              %e    a floating-point number, in scientific notation
6640              %f    a floating-point number, in fixed decimal notation
6641              %g    a floating-point number, in %e or %f notation
6642
6643           In addition, Perl permits the following widely-supported
6644           conversions:
6645
6646              %X    like %x, but using upper-case letters
6647              %E    like %e, but using an upper-case "E"
6648              %G    like %g, but with an upper-case "E" (if applicable)
6649              %b    an unsigned integer, in binary
6650              %B    like %b, but using an upper-case "B" with the # flag
6651              %p    a pointer (outputs the Perl value's address in hexadecimal)
6652              %n    special: *stores* the number of characters output so far
6653                    into the next argument in the parameter list
6654              %a    hexadecimal floating point
6655              %A    like %a, but using upper-case letters
6656
6657           Finally, for backward (and we do mean "backward") compatibility,
6658           Perl permits these unnecessary but widely-supported conversions:
6659
6660              %i    a synonym for %d
6661              %D    a synonym for %ld
6662              %U    a synonym for %lu
6663              %O    a synonym for %lo
6664              %F    a synonym for %f
6665
6666           Note that the number of exponent digits in the scientific notation
6667           produced by %e, %E, %g and %G for numbers with the modulus of the
6668           exponent less than 100 is system-dependent: it may be three or less
6669           (zero-padded as necessary).  In other words, 1.23 times ten to the
6670           99th may be either "1.23e99" or "1.23e099".  Similarly for %a and
6671           %A: the exponent or the hexadecimal digits may float: especially
6672           the "long doubles" Perl configuration option may cause surprises.
6673
6674           Between the "%" and the format letter, you may specify several
6675           additional attributes controlling the interpretation of the format.
6676           In order, these are:
6677
6678           format parameter index
6679               An explicit format parameter index, such as "2$".  By default
6680               sprintf will format the next unused argument in the list, but
6681               this allows you to take the arguments out of order:
6682
6683                 printf '%2$d %1$d', 12, 34;      # prints "34 12"
6684                 printf '%3$d %d %1$d', 1, 2, 3;  # prints "3 1 1"
6685
6686           flags
6687               one or more of:
6688
6689                  space   prefix non-negative number with a space
6690                  +       prefix non-negative number with a plus sign
6691                  -       left-justify within the field
6692                  0       use zeros, not spaces, to right-justify
6693                  #       ensure the leading "0" for any octal,
6694                          prefix non-zero hexadecimal with "0x" or "0X",
6695                          prefix non-zero binary with "0b" or "0B"
6696
6697               For example:
6698
6699                 printf '<% d>',  12;   # prints "< 12>"
6700                 printf '<% d>',   0;   # prints "< 0>"
6701                 printf '<% d>', -12;   # prints "<-12>"
6702                 printf '<%+d>',  12;   # prints "<+12>"
6703                 printf '<%+d>',   0;   # prints "<+0>"
6704                 printf '<%+d>', -12;   # prints "<-12>"
6705                 printf '<%6s>',  12;   # prints "<    12>"
6706                 printf '<%-6s>', 12;   # prints "<12    >"
6707                 printf '<%06s>', 12;   # prints "<000012>"
6708                 printf '<%#o>',  12;   # prints "<014>"
6709                 printf '<%#x>',  12;   # prints "<0xc>"
6710                 printf '<%#X>',  12;   # prints "<0XC>"
6711                 printf '<%#b>',  12;   # prints "<0b1100>"
6712                 printf '<%#B>',  12;   # prints "<0B1100>"
6713
6714               When a space and a plus sign are given as the flags at once,
6715               the space is ignored.
6716
6717                 printf '<%+ d>', 12;   # prints "<+12>"
6718                 printf '<% +d>', 12;   # prints "<+12>"
6719
6720               When the # flag and a precision are given in the %o conversion,
6721               the precision is incremented if it's necessary for the leading
6722               "0".
6723
6724                 printf '<%#.5o>', 012;      # prints "<00012>"
6725                 printf '<%#.5o>', 012345;   # prints "<012345>"
6726                 printf '<%#.0o>', 0;        # prints "<0>"
6727
6728           vector flag
6729               This flag tells Perl to interpret the supplied string as a
6730               vector of integers, one for each character in the string.  Perl
6731               applies the format to each integer in turn, then joins the
6732               resulting strings with a separator (a dot "." by default).
6733               This can be useful for displaying ordinal values of characters
6734               in arbitrary strings:
6735
6736                 printf "%vd", "AB\x{100}";           # prints "65.66.256"
6737                 printf "version is v%vd\n", $^V;     # Perl's version
6738
6739               Put an asterisk "*" before the "v" to override the string to
6740               use to separate the numbers:
6741
6742                 printf "address is %*vX\n", ":", $addr;   # IPv6 address
6743                 printf "bits are %0*v8b\n", " ", $bits;   # random bitstring
6744
6745               You can also explicitly specify the argument number to use for
6746               the join string using something like "*2$v"; for example:
6747
6748                 printf '%*4$vX %*4$vX %*4$vX',       # 3 IPv6 addresses
6749                         @addr[1..3], ":";
6750
6751           (minimum) width
6752               Arguments are usually formatted to be only as wide as required
6753               to display the given value.  You can override the width by
6754               putting a number here, or get the width from the next argument
6755               (with "*") or from a specified argument (e.g., with "*2$"):
6756
6757                printf "<%s>", "a";       # prints "<a>"
6758                printf "<%6s>", "a";      # prints "<     a>"
6759                printf "<%*s>", 6, "a";   # prints "<     a>"
6760                printf '<%*2$s>', "a", 6; # prints "<     a>"
6761                printf "<%2s>", "long";   # prints "<long>" (does not truncate)
6762
6763               If a field width obtained through "*" is negative, it has the
6764               same effect as the "-" flag: left-justification.
6765
6766           precision, or maximum width
6767               You can specify a precision (for numeric conversions) or a
6768               maximum width (for string conversions) by specifying a "."
6769               followed by a number.  For floating-point formats except "g"
6770               and "G", this specifies how many places right of the decimal
6771               point to show (the default being 6).  For example:
6772
6773                 # these examples are subject to system-specific variation
6774                 printf '<%f>', 1;    # prints "<1.000000>"
6775                 printf '<%.1f>', 1;  # prints "<1.0>"
6776                 printf '<%.0f>', 1;  # prints "<1>"
6777                 printf '<%e>', 10;   # prints "<1.000000e+01>"
6778                 printf '<%.1e>', 10; # prints "<1.0e+01>"
6779
6780               For "g" and "G", this specifies the maximum number of
6781               significant digits to show; for example:
6782
6783                 # These examples are subject to system-specific variation.
6784                 printf '<%g>', 1;        # prints "<1>"
6785                 printf '<%.10g>', 1;     # prints "<1>"
6786                 printf '<%g>', 100;      # prints "<100>"
6787                 printf '<%.1g>', 100;    # prints "<1e+02>"
6788                 printf '<%.2g>', 100.01; # prints "<1e+02>"
6789                 printf '<%.5g>', 100.01; # prints "<100.01>"
6790                 printf '<%.4g>', 100.01; # prints "<100>"
6791                 printf '<%.1g>', 0.0111; # prints "<0.01>"
6792                 printf '<%.2g>', 0.0111; # prints "<0.011>"
6793                 printf '<%.3g>', 0.0111; # prints "<0.0111>"
6794
6795               For integer conversions, specifying a precision implies that
6796               the output of the number itself should be zero-padded to this
6797               width, where the 0 flag is ignored:
6798
6799                 printf '<%.6d>', 1;      # prints "<000001>"
6800                 printf '<%+.6d>', 1;     # prints "<+000001>"
6801                 printf '<%-10.6d>', 1;   # prints "<000001    >"
6802                 printf '<%10.6d>', 1;    # prints "<    000001>"
6803                 printf '<%010.6d>', 1;   # prints "<    000001>"
6804                 printf '<%+10.6d>', 1;   # prints "<   +000001>"
6805
6806                 printf '<%.6x>', 1;      # prints "<000001>"
6807                 printf '<%#.6x>', 1;     # prints "<0x000001>"
6808                 printf '<%-10.6x>', 1;   # prints "<000001    >"
6809                 printf '<%10.6x>', 1;    # prints "<    000001>"
6810                 printf '<%010.6x>', 1;   # prints "<    000001>"
6811                 printf '<%#10.6x>', 1;   # prints "<  0x000001>"
6812
6813               For string conversions, specifying a precision truncates the
6814               string to fit the specified width:
6815
6816                 printf '<%.5s>', "truncated";   # prints "<trunc>"
6817                 printf '<%10.5s>', "truncated"; # prints "<     trunc>"
6818
6819               You can also get the precision from the next argument using
6820               ".*", or from a specified argument (e.g., with ".*2$"):
6821
6822                 printf '<%.6x>', 1;       # prints "<000001>"
6823                 printf '<%.*x>', 6, 1;    # prints "<000001>"
6824
6825                 printf '<%.*2$x>', 1, 6;  # prints "<000001>"
6826
6827                 printf '<%6.*2$x>', 1, 4; # prints "<  0001>"
6828
6829               If a precision obtained through "*" is negative, it counts as
6830               having no precision at all.
6831
6832                 printf '<%.*s>',  7, "string";   # prints "<string>"
6833                 printf '<%.*s>',  3, "string";   # prints "<str>"
6834                 printf '<%.*s>',  0, "string";   # prints "<>"
6835                 printf '<%.*s>', -1, "string";   # prints "<string>"
6836
6837                 printf '<%.*d>',  1, 0;   # prints "<0>"
6838                 printf '<%.*d>',  0, 0;   # prints "<>"
6839                 printf '<%.*d>', -1, 0;   # prints "<0>"
6840
6841           size
6842               For numeric conversions, you can specify the size to interpret
6843               the number as using "l", "h", "V", "q", "L", or "ll".  For
6844               integer conversions ("d u o x X b i D U O"), numbers are
6845               usually assumed to be whatever the default integer size is on
6846               your platform (usually 32 or 64 bits), but you can override
6847               this to use instead one of the standard C types, as supported
6848               by the compiler used to build Perl:
6849
6850                  hh          interpret integer as C type "char" or "unsigned
6851                              char" on Perl 5.14 or later
6852                  h           interpret integer as C type "short" or
6853                              "unsigned short"
6854                  j           interpret integer as C type "intmax_t" on Perl
6855                              5.14 or later; and prior to Perl 5.30, only with
6856                              a C99 compiler (unportable)
6857                  l           interpret integer as C type "long" or
6858                              "unsigned long"
6859                  q, L, or ll interpret integer as C type "long long",
6860                              "unsigned long long", or "quad" (typically
6861                              64-bit integers)
6862                  t           interpret integer as C type "ptrdiff_t" on Perl
6863                              5.14 or later
6864                  z           interpret integer as C types "size_t" or
6865                              "ssize_t" on Perl 5.14 or later
6866
6867               Note that, in general, using the "l" modifier (for example,
6868               when writing "%ld" or "%lu" instead of "%d" and "%u") is
6869               unnecessary when used from Perl code.  Moreover, it may be
6870               harmful, for example on Windows 64-bit where a long is 32-bits.
6871
6872               As of 5.14, none of these raises an exception if they are not
6873               supported on your platform.  However, if warnings are enabled,
6874               a warning of the "printf" warning class is issued on an
6875               unsupported conversion flag.  Should you instead prefer an
6876               exception, do this:
6877
6878                   use warnings FATAL => "printf";
6879
6880               If you would like to know about a version dependency before you
6881               start running the program, put something like this at its top:
6882
6883                   use v5.14;  # for hh/j/t/z/ printf modifiers
6884
6885               You can find out whether your Perl supports quads via Config:
6886
6887                   use Config;
6888                   if ($Config{use64bitint} eq "define"
6889                       || $Config{longsize} >= 8) {
6890                       print "Nice quads!\n";
6891                   }
6892
6893               For floating-point conversions ("e f g E F G"), numbers are
6894               usually assumed to be the default floating-point size on your
6895               platform (double or long double), but you can force "long
6896               double" with "q", "L", or "ll" if your platform supports them.
6897               You can find out whether your Perl supports long doubles via
6898               Config:
6899
6900                   use Config;
6901                   print "long doubles\n" if $Config{d_longdbl} eq "define";
6902
6903               You can find out whether Perl considers "long double" to be the
6904               default floating-point size to use on your platform via Config:
6905
6906                   use Config;
6907                   if ($Config{uselongdouble} eq "define") {
6908                       print "long doubles by default\n";
6909                   }
6910
6911               It can also be that long doubles and doubles are the same
6912               thing:
6913
6914                       use Config;
6915                       ($Config{doublesize} == $Config{longdblsize}) &&
6916                               print "doubles are long doubles\n";
6917
6918               The size specifier "V" has no effect for Perl code, but is
6919               supported for compatibility with XS code.  It means "use the
6920               standard size for a Perl integer or floating-point number",
6921               which is the default.
6922
6923           order of arguments
6924               Normally, "sprintf" takes the next unused argument as the value
6925               to format for each format specification.  If the format
6926               specification uses "*" to require additional arguments, these
6927               are consumed from the argument list in the order they appear in
6928               the format specification before the value to format.  Where an
6929               argument is specified by an explicit index, this does not
6930               affect the normal order for the arguments, even when the
6931               explicitly specified index would have been the next argument.
6932
6933               So:
6934
6935                   printf "<%*.*s>", $a, $b, $c;
6936
6937               uses $a for the width, $b for the precision, and $c as the
6938               value to format; while:
6939
6940                 printf '<%*1$.*s>', $a, $b;
6941
6942               would use $a for the width and precision, and $b as the value
6943               to format.
6944
6945               Here are some more examples; be aware that when using an
6946               explicit index, the "$" may need escaping:
6947
6948                printf "%2\$d %d\n",      12, 34;     # will print "34 12\n"
6949                printf "%2\$d %d %d\n",   12, 34;     # will print "34 12 34\n"
6950                printf "%3\$d %d %d\n",   12, 34, 56; # will print "56 12 34\n"
6951                printf "%2\$*3\$d %d\n",  12, 34,  3; # will print " 34 12\n"
6952                printf "%*1\$.*f\n",       4,  5, 10; # will print "5.0000\n"
6953
6954           If "use locale" (including "use locale ':not_characters'") is in
6955           effect and "POSIX::setlocale" has been called, the character used
6956           for the decimal separator in formatted floating-point numbers is
6957           affected by the "LC_NUMERIC" locale.  See perllocale and POSIX.
6958
6959       sqrt EXPR
6960       sqrt
6961           Return the positive square root of EXPR.  If EXPR is omitted, uses
6962           $_.  Works only for non-negative operands unless you've loaded the
6963           "Math::Complex" module.
6964
6965               use Math::Complex;
6966               print sqrt(-4);    # prints 2i
6967
6968       srand EXPR
6969       srand
6970           Sets and returns the random number seed for the "rand" operator.
6971
6972           The point of the function is to "seed" the "rand" function so that
6973           "rand" can produce a different sequence each time you run your
6974           program.  When called with a parameter, "srand" uses that for the
6975           seed; otherwise it (semi-)randomly chooses a seed.  In either case,
6976           starting with Perl 5.14, it returns the seed.  To signal that your
6977           code will work only on Perls of a recent vintage:
6978
6979               use v5.14;  # so srand returns the seed
6980
6981           If "srand" is not called explicitly, it is called implicitly
6982           without a parameter at the first use of the "rand" operator.
6983           However, there are a few situations where programs are likely to
6984           want to call "srand".  One is for generating predictable results,
6985           generally for testing or debugging.  There, you use "srand($seed)",
6986           with the same $seed each time.  Another case is that you may want
6987           to call "srand" after a "fork" to avoid child processes sharing the
6988           same seed value as the parent (and consequently each other).
6989
6990           Do not call "srand()" (i.e., without an argument) more than once
6991           per process.  The internal state of the random number generator
6992           should contain more entropy than can be provided by any seed, so
6993           calling "srand" again actually loses randomness.
6994
6995           Most implementations of "srand" take an integer and will silently
6996           truncate decimal numbers.  This means "srand(42)" will usually
6997           produce the same results as "srand(42.1)".  To be safe, always pass
6998           "srand" an integer.
6999
7000           A typical use of the returned seed is for a test program which has
7001           too many combinations to test comprehensively in the time available
7002           to it each run.  It can test a random subset each time, and should
7003           there be a failure, log the seed used for that run so that it can
7004           later be used to reproduce the same results.
7005
7006           "rand" is not cryptographically secure.  You should not rely on it
7007           in security-sensitive situations.  As of this writing, a number of
7008           third-party CPAN modules offer random number generators intended by
7009           their authors to be cryptographically secure, including:
7010           Data::Entropy, Crypt::Random, Math::Random::Secure, and
7011           Math::TrulyRandom.
7012
7013       stat FILEHANDLE
7014       stat EXPR
7015       stat DIRHANDLE
7016       stat
7017           Returns a 13-element list giving the status info for a file, either
7018           the file opened via FILEHANDLE or DIRHANDLE, or named by EXPR.  If
7019           EXPR is omitted, it stats $_ (not "_"!).  Returns the empty list if
7020           "stat" fails.  Typically used as follows:
7021
7022               my ($dev,$ino,$mode,$nlink,$uid,$gid,$rdev,$size,
7023                   $atime,$mtime,$ctime,$blksize,$blocks)
7024                      = stat($filename);
7025
7026           Not all fields are supported on all filesystem types.  Here are the
7027           meanings of the fields:
7028
7029             0 dev      device number of filesystem
7030             1 ino      inode number
7031             2 mode     file mode  (type and permissions)
7032             3 nlink    number of (hard) links to the file
7033             4 uid      numeric user ID of file's owner
7034             5 gid      numeric group ID of file's owner
7035             6 rdev     the device identifier (special files only)
7036             7 size     total size of file, in bytes
7037             8 atime    last access time in seconds since the epoch
7038             9 mtime    last modify time in seconds since the epoch
7039            10 ctime    inode change time in seconds since the epoch (*)
7040            11 blksize  preferred I/O size in bytes for interacting with the
7041                        file (may vary from file to file)
7042            12 blocks   actual number of system-specific blocks allocated
7043                        on disk (often, but not always, 512 bytes each)
7044
7045           (The epoch was at 00:00 January 1, 1970 GMT.)
7046
7047           (*) Not all fields are supported on all filesystem types.  Notably,
7048           the ctime field is non-portable.  In particular, you cannot expect
7049           it to be a "creation time"; see "Files and Filesystems" in perlport
7050           for details.
7051
7052           If "stat" is passed the special filehandle consisting of an
7053           underline, no stat is done, but the current contents of the stat
7054           structure from the last "stat", "lstat", or filetest are returned.
7055           Example:
7056
7057               if (-x $file && (($d) = stat(_)) && $d < 0) {
7058                   print "$file is executable NFS file\n";
7059               }
7060
7061           (This works on machines only for which the device number is
7062           negative under NFS.)
7063
7064           On some platforms inode numbers are of a type larger than perl
7065           knows how to handle as integer numerical values.  If necessary, an
7066           inode number will be returned as a decimal string in order to
7067           preserve the entire value.  If used in a numeric context, this will
7068           be converted to a floating-point numerical value, with rounding, a
7069           fate that is best avoided.  Therefore, you should prefer to compare
7070           inode numbers using "eq" rather than "==".  "eq" will work fine on
7071           inode numbers that are represented numerically, as well as those
7072           represented as strings.
7073
7074           Because the mode contains both the file type and its permissions,
7075           you should mask off the file type portion and (s)printf using a
7076           "%o" if you want to see the real permissions.
7077
7078               my $mode = (stat($filename))[2];
7079               printf "Permissions are %04o\n", $mode & 07777;
7080
7081           In scalar context, "stat" returns a boolean value indicating
7082           success or failure, and, if successful, sets the information
7083           associated with the special filehandle "_".
7084
7085           The File::stat module provides a convenient, by-name access
7086           mechanism:
7087
7088               use File::stat;
7089               my $sb = stat($filename);
7090               printf "File is %s, size is %s, perm %04o, mtime %s\n",
7091                      $filename, $sb->size, $sb->mode & 07777,
7092                      scalar localtime $sb->mtime;
7093
7094           You can import symbolic mode constants ("S_IF*") and functions
7095           ("S_IS*") from the Fcntl module:
7096
7097               use Fcntl ':mode';
7098
7099               my $mode = (stat($filename))[2];
7100
7101               my $user_rwx      = ($mode & S_IRWXU) >> 6;
7102               my $group_read    = ($mode & S_IRGRP) >> 3;
7103               my $other_execute =  $mode & S_IXOTH;
7104
7105               printf "Permissions are %04o\n", S_IMODE($mode), "\n";
7106
7107               my $is_setuid     =  $mode & S_ISUID;
7108               my $is_directory  =  S_ISDIR($mode);
7109
7110           You could write the last two using the "-u" and "-d" operators.
7111           Commonly available "S_IF*" constants are:
7112
7113               # Permissions: read, write, execute, for user, group, others.
7114
7115               S_IRWXU S_IRUSR S_IWUSR S_IXUSR
7116               S_IRWXG S_IRGRP S_IWGRP S_IXGRP
7117               S_IRWXO S_IROTH S_IWOTH S_IXOTH
7118
7119               # Setuid/Setgid/Stickiness/SaveText.
7120               # Note that the exact meaning of these is system-dependent.
7121
7122               S_ISUID S_ISGID S_ISVTX S_ISTXT
7123
7124               # File types.  Not all are necessarily available on
7125               # your system.
7126
7127               S_IFREG S_IFDIR S_IFLNK S_IFBLK S_IFCHR
7128               S_IFIFO S_IFSOCK S_IFWHT S_ENFMT
7129
7130               # The following are compatibility aliases for S_IRUSR,
7131               # S_IWUSR, and S_IXUSR.
7132
7133               S_IREAD S_IWRITE S_IEXEC
7134
7135           and the "S_IF*" functions are
7136
7137               S_IMODE($mode)    the part of $mode containing the permission
7138                                 bits and the setuid/setgid/sticky bits
7139
7140               S_IFMT($mode)     the part of $mode containing the file type
7141                                 which can be bit-anded with (for example)
7142                                 S_IFREG or with the following functions
7143
7144               # The operators -f, -d, -l, -b, -c, -p, and -S.
7145
7146               S_ISREG($mode) S_ISDIR($mode) S_ISLNK($mode)
7147               S_ISBLK($mode) S_ISCHR($mode) S_ISFIFO($mode) S_ISSOCK($mode)
7148
7149               # No direct -X operator counterpart, but for the first one
7150               # the -g operator is often equivalent.  The ENFMT stands for
7151               # record flocking enforcement, a platform-dependent feature.
7152
7153               S_ISENFMT($mode) S_ISWHT($mode)
7154
7155           See your native chmod(2) and stat(2) documentation for more details
7156           about the "S_*" constants.  To get status info for a symbolic link
7157           instead of the target file behind the link, use the "lstat"
7158           function.
7159
7160           Portability issues: "stat" in perlport.
7161
7162       state VARLIST
7163       state TYPE VARLIST
7164       state VARLIST : ATTRS
7165       state TYPE VARLIST : ATTRS
7166           "state" declares a lexically scoped variable, just like "my".
7167           However, those variables will never be reinitialized, contrary to
7168           lexical variables that are reinitialized each time their enclosing
7169           block is entered.  See "Persistent Private Variables" in perlsub
7170           for details.
7171
7172           If more than one variable is listed, the list must be placed in
7173           parentheses.  With a parenthesised list, "undef" can be used as a
7174           dummy placeholder.  However, since initialization of state
7175           variables in such lists is currently not possible this would serve
7176           no purpose.
7177
7178           Redeclaring a variable in the same scope or statement will "shadow"
7179           the previous declaration, creating a new instance and preventing
7180           access to the previous one. This is usually undesired and, if
7181           warnings are enabled, will result in a warning in the "shadow"
7182           category.
7183
7184           "state" is available only if the "state" feature is enabled or if
7185           it is prefixed with "CORE::".  The "state" feature is enabled
7186           automatically with a "use v5.10" (or higher) declaration in the
7187           current scope.
7188
7189       study SCALAR
7190       study
7191           At this time, "study" does nothing. This may change in the future.
7192
7193           Prior to Perl version 5.16, it would create an inverted index of
7194           all characters that occurred in the given SCALAR (or $_ if
7195           unspecified). When matching a pattern, the rarest character from
7196           the pattern would be looked up in this index. Rarity was based on
7197           some static frequency tables constructed from some C programs and
7198           English text.
7199
7200       sub NAME BLOCK
7201       sub NAME (PROTO) BLOCK
7202       sub NAME : ATTRS BLOCK
7203       sub NAME (PROTO) : ATTRS BLOCK
7204           This is subroutine definition, not a real function per se.  Without
7205           a BLOCK it's just a forward declaration.  Without a NAME, it's an
7206           anonymous function declaration, so does return a value: the CODE
7207           ref of the closure just created.
7208
7209           See perlsub and perlref for details about subroutines and
7210           references; see attributes and Attribute::Handlers for more
7211           information about attributes.
7212
7213       __SUB__
7214           A special token that returns a reference to the current subroutine,
7215           or "undef" outside of a subroutine.
7216
7217           The behaviour of "__SUB__" within a regex code block (such as
7218           "/(?{...})/") is subject to change.
7219
7220           This token is only available under "use v5.16" or the "current_sub"
7221           feature.  See feature.
7222
7223       substr EXPR,OFFSET,LENGTH,REPLACEMENT
7224       substr EXPR,OFFSET,LENGTH
7225       substr EXPR,OFFSET
7226           Extracts a substring out of EXPR and returns it.  First character
7227           is at offset zero.  If OFFSET is negative, starts that far back
7228           from the end of the string.  If LENGTH is omitted, returns
7229           everything through the end of the string.  If LENGTH is negative,
7230           leaves that many characters off the end of the string.
7231
7232               my $s = "The black cat climbed the green tree";
7233               my $color  = substr $s, 4, 5;      # black
7234               my $middle = substr $s, 4, -11;    # black cat climbed the
7235               my $end    = substr $s, 14;        # climbed the green tree
7236               my $tail   = substr $s, -4;        # tree
7237               my $z      = substr $s, -4, 2;     # tr
7238
7239           You can use the "substr" function as an lvalue, in which case EXPR
7240           must itself be an lvalue.  If you assign something shorter than
7241           LENGTH, the string will shrink, and if you assign something longer
7242           than LENGTH, the string will grow to accommodate it.  To keep the
7243           string the same length, you may need to pad or chop your value
7244           using "sprintf".
7245
7246           If OFFSET and LENGTH specify a substring that is partly outside the
7247           string, only the part within the string is returned.  If the
7248           substring is beyond either end of the string, "substr" returns the
7249           undefined value and produces a warning.  When used as an lvalue,
7250           specifying a substring that is entirely outside the string raises
7251           an exception.  Here's an example showing the behavior for boundary
7252           cases:
7253
7254               my $name = 'fred';
7255               substr($name, 4) = 'dy';         # $name is now 'freddy'
7256               my $null = substr $name, 6, 2;   # returns "" (no warning)
7257               my $oops = substr $name, 7;      # returns undef, with warning
7258               substr($name, 7) = 'gap';        # raises an exception
7259
7260           An alternative to using "substr" as an lvalue is to specify the
7261           replacement string as the 4th argument.  This allows you to replace
7262           parts of the EXPR and return what was there before in one
7263           operation, just as you can with "splice".
7264
7265               my $s = "The black cat climbed the green tree";
7266               my $z = substr $s, 14, 7, "jumped from";    # climbed
7267               # $s is now "The black cat jumped from the green tree"
7268
7269           Note that the lvalue returned by the three-argument version of
7270           "substr" acts as a 'magic bullet'; each time it is assigned to, it
7271           remembers which part of the original string is being modified; for
7272           example:
7273
7274               my $x = '1234';
7275               for (substr($x,1,2)) {
7276                   $_ = 'a';   print $x,"\n";    # prints 1a4
7277                   $_ = 'xyz'; print $x,"\n";    # prints 1xyz4
7278                   $x = '56789';
7279                   $_ = 'pq';  print $x,"\n";    # prints 5pq9
7280               }
7281
7282           With negative offsets, it remembers its position from the end of
7283           the string when the target string is modified:
7284
7285               my $x = '1234';
7286               for (substr($x, -3, 2)) {
7287                   $_ = 'a';   print $x,"\n";    # prints 1a4, as above
7288                   $x = 'abcdefg';
7289                   print $_,"\n";                # prints f
7290               }
7291
7292           Prior to Perl version 5.10, the result of using an lvalue multiple
7293           times was unspecified.  Prior to 5.16, the result with negative
7294           offsets was unspecified.
7295
7296       symlink OLDFILE,NEWFILE
7297           Creates a new filename symbolically linked to the old filename.
7298           Returns 1 for success, 0 otherwise.  On systems that don't support
7299           symbolic links, raises an exception.  To check for that, use eval:
7300
7301               my $symlink_exists = eval { symlink("",""); 1 };
7302
7303           Portability issues: "symlink" in perlport.
7304
7305       syscall NUMBER, LIST
7306           Calls the system call specified as the first element of the list,
7307           passing the remaining elements as arguments to the system call.  If
7308           unimplemented, raises an exception.  The arguments are interpreted
7309           as follows: if a given argument is numeric, the argument is passed
7310           as an int.  If not, the pointer to the string value is passed.  You
7311           are responsible to make sure a string is pre-extended long enough
7312           to receive any result that might be written into a string.  You
7313           can't use a string literal (or other read-only string) as an
7314           argument to "syscall" because Perl has to assume that any string
7315           pointer might be written through.  If your integer arguments are
7316           not literals and have never been interpreted in a numeric context,
7317           you may need to add 0 to them to force them to look like numbers.
7318           This emulates the "syswrite" function (or vice versa):
7319
7320               require 'syscall.ph';        # may need to run h2ph
7321               my $s = "hi there\n";
7322               syscall(SYS_write(), fileno(STDOUT), $s, length $s);
7323
7324           Note that Perl supports passing of up to only 14 arguments to your
7325           syscall, which in practice should (usually) suffice.
7326
7327           Syscall returns whatever value returned by the system call it
7328           calls.  If the system call fails, "syscall" returns "-1" and sets
7329           $! (errno).  Note that some system calls can legitimately return
7330           "-1".  The proper way to handle such calls is to assign "$! = 0"
7331           before the call, then check the value of $! if "syscall" returns
7332           "-1".
7333
7334           There's a problem with "syscall(SYS_pipe())": it returns the file
7335           number of the read end of the pipe it creates, but there is no way
7336           to retrieve the file number of the other end.  You can avoid this
7337           problem by using "pipe" instead.
7338
7339           Portability issues: "syscall" in perlport.
7340
7341       sysopen FILEHANDLE,FILENAME,MODE
7342       sysopen FILEHANDLE,FILENAME,MODE,PERMS
7343           Opens the file whose filename is given by FILENAME, and associates
7344           it with FILEHANDLE.  If FILEHANDLE is an expression, its value is
7345           used as the real filehandle wanted; an undefined scalar will be
7346           suitably autovivified.  This function calls the underlying
7347           operating system's open(2) function with the parameters FILENAME,
7348           MODE, and PERMS.
7349
7350           Returns true on success and "undef" otherwise.
7351
7352           PerlIO layers will be applied to the handle the same way they would
7353           in an "open" call that does not specify layers. That is, the
7354           current value of "${^OPEN}" as set by the open pragma in a lexical
7355           scope, or the "-C" commandline option or "PERL_UNICODE" environment
7356           variable in the main program scope, falling back to the platform
7357           defaults as described in "Defaults and how to override them" in
7358           PerlIO. If you want to remove any layers that may transform the
7359           byte stream, use "binmode" after opening it.
7360
7361           The possible values and flag bits of the MODE parameter are system-
7362           dependent; they are available via the standard module "Fcntl".  See
7363           the documentation of your operating system's open(2) syscall to see
7364           which values and flag bits are available.  You may combine several
7365           flags using the "|"-operator.
7366
7367           Some of the most common values are "O_RDONLY" for opening the file
7368           in read-only mode, "O_WRONLY" for opening the file in write-only
7369           mode, and "O_RDWR" for opening the file in read-write mode.
7370
7371           For historical reasons, some values work on almost every system
7372           supported by Perl: 0 means read-only, 1 means write-only, and 2
7373           means read/write.  We know that these values do not work under
7374           OS/390; you probably don't want to use them in new code.
7375
7376           If the file named by FILENAME does not exist and the "open" call
7377           creates it (typically because MODE includes the "O_CREAT" flag),
7378           then the value of PERMS specifies the permissions of the newly
7379           created file.  If you omit the PERMS argument to "sysopen", Perl
7380           uses the octal value 0666.  These permission values need to be in
7381           octal, and are modified by your process's current "umask".
7382
7383           In many systems the "O_EXCL" flag is available for opening files in
7384           exclusive mode.  This is not locking: exclusiveness means here that
7385           if the file already exists, "sysopen" fails.  "O_EXCL" may not work
7386           on network filesystems, and has no effect unless the "O_CREAT" flag
7387           is set as well.  Setting "O_CREAT|O_EXCL" prevents the file from
7388           being opened if it is a symbolic link.  It does not protect against
7389           symbolic links in the file's path.
7390
7391           Sometimes you may want to truncate an already-existing file.  This
7392           can be done using the "O_TRUNC" flag.  The behavior of "O_TRUNC"
7393           with "O_RDONLY" is undefined.
7394
7395           You should seldom if ever use 0644 as argument to "sysopen",
7396           because that takes away the user's option to have a more permissive
7397           umask.  Better to omit it.  See "umask" for more on this.
7398
7399           This function has no direct relation to the usage of "sysread",
7400           "syswrite", or "sysseek".  A handle opened with this function can
7401           be used with buffered IO just as one opened with "open" can be used
7402           with unbuffered IO.
7403
7404           Note that under Perls older than 5.8.0, "sysopen" depends on the
7405           fdopen(3) C library function.  On many Unix systems, fdopen(3) is
7406           known to fail when file descriptors exceed a certain value,
7407           typically 255.  If you need more file descriptors than that,
7408           consider using the "POSIX::open" function.  For Perls 5.8.0 and
7409           later, PerlIO is (most often) the default.
7410
7411           See perlopentut for a kinder, gentler explanation of opening files.
7412
7413           Portability issues: "sysopen" in perlport.
7414
7415       sysread FILEHANDLE,SCALAR,LENGTH,OFFSET
7416       sysread FILEHANDLE,SCALAR,LENGTH
7417           Attempts to read LENGTH bytes of data into variable SCALAR from the
7418           specified FILEHANDLE, using read(2).  It bypasses any PerlIO layers
7419           including buffered IO (but is affected by the presence of the
7420           ":utf8" layer as described later), so mixing this with other kinds
7421           of reads, "print", "write", "seek", "tell", or "eof" can cause
7422           confusion because the ":perlio" or ":crlf" layers usually buffer
7423           data.  Returns the number of bytes actually read, 0 at end of file,
7424           or undef if there was an error (in the latter case $! is also set).
7425           SCALAR will be grown or shrunk so that the last byte actually read
7426           is the last byte of the scalar after the read.
7427
7428           An OFFSET may be specified to place the read data at some place in
7429           the string other than the beginning.  A negative OFFSET specifies
7430           placement at that many characters counting backwards from the end
7431           of the string.  A positive OFFSET greater than the length of SCALAR
7432           results in the string being padded to the required size with "\0"
7433           bytes before the result of the read is appended.
7434
7435           There is no syseof() function, which is ok, since "eof" doesn't
7436           work well on device files (like ttys) anyway.  Use "sysread" and
7437           check for a return value of 0 to decide whether you're done.
7438
7439           Note that if the filehandle has been marked as ":utf8", "sysread"
7440           will throw an exception.  The ":encoding(...)" layer implicitly
7441           introduces the ":utf8" layer.  See "binmode", "open", and the open
7442           pragma.
7443
7444       sysseek FILEHANDLE,POSITION,WHENCE
7445           Sets FILEHANDLE's system position in bytes using lseek(2).
7446           FILEHANDLE may be an expression whose value gives the name of the
7447           filehandle.  The values for WHENCE are 0 to set the new position to
7448           POSITION; 1 to set it to the current position plus POSITION; and 2
7449           to set it to EOF plus POSITION, typically negative.
7450
7451           Note the emphasis on bytes: even if the filehandle has been set to
7452           operate on characters (for example using the ":encoding(UTF-8)" I/O
7453           layer), the "seek", "tell", and "sysseek" family of functions use
7454           byte offsets, not character offsets, because seeking to a character
7455           offset would be very slow in a UTF-8 file.
7456
7457           "sysseek" bypasses normal buffered IO, so mixing it with reads
7458           other than "sysread" (for example "readline" or "read"), "print",
7459           "write", "seek", "tell", or "eof" may cause confusion.
7460
7461           For WHENCE, you may also use the constants "SEEK_SET", "SEEK_CUR",
7462           and "SEEK_END" (start of the file, current position, end of the
7463           file) from the Fcntl module.  Use of the constants is also more
7464           portable than relying on 0, 1, and 2.  For example to define a
7465           "systell" function:
7466
7467               use Fcntl 'SEEK_CUR';
7468               sub systell { sysseek($_[0], 0, SEEK_CUR) }
7469
7470           Returns the new position, or the undefined value on failure.  A
7471           position of zero is returned as the string "0 but true"; thus
7472           "sysseek" returns true on success and false on failure, yet you can
7473           still easily determine the new position.
7474
7475       system LIST
7476       system PROGRAM LIST
7477           Does exactly the same thing as "exec", except that a fork is done
7478           first and the parent process waits for the child process to exit.
7479           Note that argument processing varies depending on the number of
7480           arguments.  If there is more than one argument in LIST, or if LIST
7481           is an array with more than one value, starts the program given by
7482           the first element of the list with arguments given by the rest of
7483           the list.  If there is only one scalar argument, the argument is
7484           checked for shell metacharacters, and if there are any, the entire
7485           argument is passed to the system's command shell for parsing (this
7486           is "/bin/sh -c" on Unix platforms, but varies on other platforms).
7487           If there are no shell metacharacters in the argument, it is split
7488           into words and passed directly to "execvp", which is more
7489           efficient.  On Windows, only the "system PROGRAM LIST" syntax will
7490           reliably avoid using the shell; "system LIST", even with more than
7491           one element, will fall back to the shell if the first spawn fails.
7492
7493           Perl will attempt to flush all files opened for output before any
7494           operation that may do a fork, but this may not be supported on some
7495           platforms (see perlport).  To be safe, you may need to set $|
7496           ($AUTOFLUSH in English) or call the "autoflush" method of
7497           "IO::Handle" on any open handles.
7498
7499           The return value is the exit status of the program as returned by
7500           the "wait" call.  To get the actual exit value, shift right by
7501           eight (see below).  See also "exec".  This is not what you want to
7502           use to capture the output from a command; for that you should use
7503           merely backticks or "qx//", as described in "`STRING`" in perlop.
7504           Return value of -1 indicates a failure to start the program or an
7505           error of the wait(2) system call (inspect $! for the reason).
7506
7507           If you'd like to make "system" (and many other bits of Perl) die on
7508           error, have a look at the autodie pragma.
7509
7510           Like "exec", "system" allows you to lie to a program about its name
7511           if you use the "system PROGRAM LIST" syntax.  Again, see "exec".
7512
7513           Since "SIGINT" and "SIGQUIT" are ignored during the execution of
7514           "system", if you expect your program to terminate on receipt of
7515           these signals you will need to arrange to do so yourself based on
7516           the return value.
7517
7518               my @args = ("command", "arg1", "arg2");
7519               system(@args) == 0
7520                   or die "system @args failed: $?";
7521
7522           If you'd like to manually inspect "system"'s failure, you can check
7523           all possible failure modes by inspecting $? like this:
7524
7525               if ($? == -1) {
7526                   print "failed to execute: $!\n";
7527               }
7528               elsif ($? & 127) {
7529                   printf "child died with signal %d, %s coredump\n",
7530                       ($? & 127),  ($? & 128) ? 'with' : 'without';
7531               }
7532               else {
7533                   printf "child exited with value %d\n", $? >> 8;
7534               }
7535
7536           Alternatively, you may inspect the value of
7537           "${^CHILD_ERROR_NATIVE}" with the "W*()" calls from the POSIX
7538           module.
7539
7540           When "system"'s arguments are executed indirectly by the shell,
7541           results and return codes are subject to its quirks.  See "`STRING`"
7542           in perlop and "exec" for details.
7543
7544           Since "system" does a "fork" and "wait" it may affect a "SIGCHLD"
7545           handler.  See perlipc for details.
7546
7547           Portability issues: "system" in perlport.
7548
7549       syswrite FILEHANDLE,SCALAR,LENGTH,OFFSET
7550       syswrite FILEHANDLE,SCALAR,LENGTH
7551       syswrite FILEHANDLE,SCALAR
7552           Attempts to write LENGTH bytes of data from variable SCALAR to the
7553           specified FILEHANDLE, using write(2).  If LENGTH is not specified,
7554           writes whole SCALAR.  It bypasses any PerlIO layers including
7555           buffered IO (but is affected by the presence of the ":utf8" layer
7556           as described later), so mixing this with reads (other than
7557           "sysread)"), "print", "write", "seek", "tell", or "eof" may cause
7558           confusion because the ":perlio" and ":crlf" layers usually buffer
7559           data.  Returns the number of bytes actually written, or "undef" if
7560           there was an error (in this case the errno variable $! is also
7561           set).  If the LENGTH is greater than the data available in the
7562           SCALAR after the OFFSET, only as much data as is available will be
7563           written.
7564
7565           An OFFSET may be specified to write the data from some part of the
7566           string other than the beginning.  A negative OFFSET specifies
7567           writing that many characters counting backwards from the end of the
7568           string.  If SCALAR is of length zero, you can only use an OFFSET of
7569           0.
7570
7571           WARNING: If the filehandle is marked ":utf8", "syswrite" will raise
7572           an exception.  The ":encoding(...)" layer implicitly introduces the
7573           ":utf8" layer.  Alternately, if the handle is not marked with an
7574           encoding but you attempt to write characters with code points over
7575           255, raises an exception.  See "binmode", "open", and the open
7576           pragma.
7577
7578       tell FILEHANDLE
7579       tell
7580           Returns the current position in bytes for FILEHANDLE, or -1 on
7581           error.  FILEHANDLE may be an expression whose value gives the name
7582           of the actual filehandle.  If FILEHANDLE is omitted, assumes the
7583           file last read.
7584
7585           Note the emphasis on bytes: even if the filehandle has been set to
7586           operate on characters (for example using the ":encoding(UTF-8)" I/O
7587           layer), the "seek", "tell", and "sysseek" family of functions use
7588           byte offsets, not character offsets, because seeking to a character
7589           offset would be very slow in a UTF-8 file.
7590
7591           The return value of "tell" for the standard streams like the STDIN
7592           depends on the operating system: it may return -1 or something
7593           else.  "tell" on pipes, fifos, and sockets usually returns -1.
7594
7595           There is no "systell" function.  Use "sysseek($fh, 0, 1)" for that.
7596
7597           Do not use "tell" (or other buffered I/O operations) on a
7598           filehandle that has been manipulated by "sysread", "syswrite", or
7599           "sysseek".  Those functions ignore the buffering, while "tell" does
7600           not.
7601
7602       telldir DIRHANDLE
7603           Returns the current position of the "readdir" routines on
7604           DIRHANDLE.  Value may be given to "seekdir" to access a particular
7605           location in a directory.  "telldir" has the same caveats about
7606           possible directory compaction as the corresponding system library
7607           routine.
7608
7609       tie VARIABLE,CLASSNAME,LIST
7610           This function binds a variable to a package class that will provide
7611           the implementation for the variable.  VARIABLE is the name of the
7612           variable to be enchanted.  CLASSNAME is the name of a class
7613           implementing objects of correct type.  Any additional arguments are
7614           passed to the appropriate constructor method of the class (meaning
7615           "TIESCALAR", "TIEHANDLE", "TIEARRAY", or "TIEHASH").  Typically
7616           these are arguments such as might be passed to the dbm_open(3)
7617           function of C.  The object returned by the constructor is also
7618           returned by the "tie" function, which would be useful if you want
7619           to access other methods in CLASSNAME.
7620
7621           Note that functions such as "keys" and "values" may return huge
7622           lists when used on large objects, like DBM files.  You may prefer
7623           to use the "each" function to iterate over such.  Example:
7624
7625               # print out history file offsets
7626               use NDBM_File;
7627               tie(my %HIST, 'NDBM_File', '/usr/lib/news/history', 1, 0);
7628               while (my ($key,$val) = each %HIST) {
7629                   print $key, ' = ', unpack('L', $val), "\n";
7630               }
7631
7632           A class implementing a hash should have the following methods:
7633
7634               TIEHASH classname, LIST
7635               FETCH this, key
7636               STORE this, key, value
7637               DELETE this, key
7638               CLEAR this
7639               EXISTS this, key
7640               FIRSTKEY this
7641               NEXTKEY this, lastkey
7642               SCALAR this
7643               DESTROY this
7644               UNTIE this
7645
7646           A class implementing an ordinary array should have the following
7647           methods:
7648
7649               TIEARRAY classname, LIST
7650               FETCH this, key
7651               STORE this, key, value
7652               FETCHSIZE this
7653               STORESIZE this, count
7654               CLEAR this
7655               PUSH this, LIST
7656               POP this
7657               SHIFT this
7658               UNSHIFT this, LIST
7659               SPLICE this, offset, length, LIST
7660               EXTEND this, count
7661               DELETE this, key
7662               EXISTS this, key
7663               DESTROY this
7664               UNTIE this
7665
7666           A class implementing a filehandle should have the following
7667           methods:
7668
7669               TIEHANDLE classname, LIST
7670               READ this, scalar, length, offset
7671               READLINE this
7672               GETC this
7673               WRITE this, scalar, length, offset
7674               PRINT this, LIST
7675               PRINTF this, format, LIST
7676               BINMODE this
7677               EOF this
7678               FILENO this
7679               SEEK this, position, whence
7680               TELL this
7681               OPEN this, mode, LIST
7682               CLOSE this
7683               DESTROY this
7684               UNTIE this
7685
7686           A class implementing a scalar should have the following methods:
7687
7688               TIESCALAR classname, LIST
7689               FETCH this,
7690               STORE this, value
7691               DESTROY this
7692               UNTIE this
7693
7694           Not all methods indicated above need be implemented.  See perltie,
7695           Tie::Hash, Tie::Array, Tie::Scalar, and Tie::Handle.
7696
7697           Unlike "dbmopen", the "tie" function will not "use" or "require" a
7698           module for you; you need to do that explicitly yourself.  See
7699           DB_File or the Config module for interesting "tie" implementations.
7700
7701           For further details see perltie, "tied".
7702
7703       tied VARIABLE
7704           Returns a reference to the object underlying VARIABLE (the same
7705           value that was originally returned by the "tie" call that bound the
7706           variable to a package.)  Returns the undefined value if VARIABLE
7707           isn't tied to a package.
7708
7709       time
7710           Returns the number of non-leap seconds since whatever time the
7711           system considers to be the epoch, suitable for feeding to "gmtime"
7712           and "localtime".  On most systems the epoch is 00:00:00 UTC,
7713           January 1, 1970; a prominent exception being Mac OS Classic which
7714           uses 00:00:00, January 1, 1904 in the current local time zone for
7715           its epoch.
7716
7717           For measuring time in better granularity than one second, use the
7718           Time::HiRes module from Perl 5.8 onwards (or from CPAN before
7719           then), or, if you have gettimeofday(2), you may be able to use the
7720           "syscall" interface of Perl.  See perlfaq8 for details.
7721
7722           For date and time processing look at the many related modules on
7723           CPAN.  For a comprehensive date and time representation look at the
7724           DateTime module.
7725
7726       times
7727           Returns a four-element list giving the user and system times in
7728           seconds for this process and any exited children of this process.
7729
7730               my ($user,$system,$cuser,$csystem) = times;
7731
7732           In scalar context, "times" returns $user.
7733
7734           Children's times are only included for terminated children.
7735
7736           Portability issues: "times" in perlport.
7737
7738       tr///
7739           The transliteration operator.  Same as "y///".  See "Quote-Like
7740           Operators" in perlop.
7741
7742       truncate FILEHANDLE,LENGTH
7743       truncate EXPR,LENGTH
7744           Truncates the file opened on FILEHANDLE, or named by EXPR, to the
7745           specified length.  Raises an exception if truncate isn't
7746           implemented on your system.  Returns true if successful, "undef" on
7747           error.
7748
7749           The behavior is undefined if LENGTH is greater than the length of
7750           the file.
7751
7752           The position in the file of FILEHANDLE is left unchanged.  You may
7753           want to call seek before writing to the file.
7754
7755           Portability issues: "truncate" in perlport.
7756
7757       uc EXPR
7758       uc  Returns an uppercased version of EXPR.  If EXPR is omitted, uses
7759           $_.
7760
7761               my $str = uc("Perl is GREAT"); # "PERL IS GREAT"
7762
7763           This function behaves the same way under various pragmas, such as
7764           in a locale, as "lc" does.
7765
7766           If you want titlecase mapping on initial letters see "ucfirst"
7767           instead.
7768
7769           Note: This is the internal function implementing the "\U" escape in
7770           double-quoted strings.
7771
7772               my $str = "Perl is \Ugreat\E"; # "Perl is GREAT"
7773
7774       ucfirst EXPR
7775       ucfirst
7776           Returns the value of EXPR with the first character in uppercase
7777           (titlecase in Unicode).  This is the internal function implementing
7778           the "\u" escape in double-quoted strings.
7779
7780           If EXPR is omitted, uses $_.
7781
7782           This function behaves the same way under various pragmas, such as
7783           in a locale, as "lc" does.
7784
7785       umask EXPR
7786       umask
7787           Sets the umask for the process to EXPR and returns the previous
7788           value.  If EXPR is omitted, merely returns the current umask.
7789
7790           The Unix permission "rwxr-x---" is represented as three sets of
7791           three bits, or three octal digits: 0750 (the leading 0 indicates
7792           octal and isn't one of the digits).  The "umask" value is such a
7793           number representing disabled permissions bits.  The permission (or
7794           "mode") values you pass "mkdir" or "sysopen" are modified by your
7795           umask, so even if you tell "sysopen" to create a file with
7796           permissions 0777, if your umask is 0022, then the file will
7797           actually be created with permissions 0755.  If your "umask" were
7798           0027 (group can't write; others can't read, write, or execute),
7799           then passing "sysopen" 0666 would create a file with mode 0640
7800           (because "0666 &~ 027" is 0640).
7801
7802           Here's some advice: supply a creation mode of 0666 for regular
7803           files (in "sysopen") and one of 0777 for directories (in "mkdir")
7804           and executable files.  This gives users the freedom of choice: if
7805           they want protected files, they might choose process umasks of 022,
7806           027, or even the particularly antisocial mask of 077.  Programs
7807           should rarely if ever make policy decisions better left to the
7808           user.  The exception to this is when writing files that should be
7809           kept private: mail files, web browser cookies, .rhosts files, and
7810           so on.
7811
7812           If umask(2) is not implemented on your system and you are trying to
7813           restrict access for yourself (i.e., "(EXPR & 0700) > 0"), raises an
7814           exception.  If umask(2) is not implemented and you are not trying
7815           to restrict access for yourself, returns "undef".
7816
7817           Remember that a umask is a number, usually given in octal; it is
7818           not a string of octal digits.  See also "oct", if all you have is a
7819           string.
7820
7821           Portability issues: "umask" in perlport.
7822
7823       undef EXPR
7824       undef
7825           Undefines the value of EXPR, which must be an lvalue.  Use only on
7826           a scalar value, an array (using "@"), a hash (using "%"), a
7827           subroutine (using "&"), or a typeglob (using "*").  Saying "undef
7828           $hash{$key}" will probably not do what you expect on most
7829           predefined variables or DBM list values, so don't do that; see
7830           "delete".  Always returns the undefined value.  You can omit the
7831           EXPR, in which case nothing is undefined, but you still get an
7832           undefined value that you could, for instance, return from a
7833           subroutine, assign to a variable, or pass as a parameter.
7834           Examples:
7835
7836               undef $foo;
7837               undef $bar{'blurfl'};      # Compare to: delete $bar{'blurfl'};
7838               undef @ary;
7839               undef %hash;
7840               undef &mysub;
7841               undef *xyz;       # destroys $xyz, @xyz, %xyz, &xyz, etc.
7842               return (wantarray ? (undef, $errmsg) : undef) if $they_blew_it;
7843               select undef, undef, undef, 0.25;
7844               my ($x, $y, undef, $z) = foo();    # Ignore third value returned
7845
7846           Note that this is a unary operator, not a list operator.
7847
7848       unlink LIST
7849       unlink
7850           Deletes a list of files.  On success, it returns the number of
7851           files it successfully deleted.  On failure, it returns false and
7852           sets $! (errno):
7853
7854               my $unlinked = unlink 'a', 'b', 'c';
7855               unlink @goners;
7856               unlink glob "*.bak";
7857
7858           On error, "unlink" will not tell you which files it could not
7859           remove.  If you want to know which files you could not remove, try
7860           them one at a time:
7861
7862                foreach my $file ( @goners ) {
7863                    unlink $file or warn "Could not unlink $file: $!";
7864                }
7865
7866           Note: "unlink" will not attempt to delete directories unless you
7867           are superuser and the -U flag is supplied to Perl.  Even if these
7868           conditions are met, be warned that unlinking a directory can
7869           inflict damage on your filesystem.  Finally, using "unlink" on
7870           directories is not supported on many operating systems.  Use
7871           "rmdir" instead.
7872
7873           If LIST is omitted, "unlink" uses $_.
7874
7875       unpack TEMPLATE,EXPR
7876       unpack TEMPLATE
7877           "unpack" does the reverse of "pack": it takes a string and expands
7878           it out into a list of values.  (In scalar context, it returns
7879           merely the first value produced.)
7880
7881           If EXPR is omitted, unpacks the $_ string.  See perlpacktut for an
7882           introduction to this function.
7883
7884           The string is broken into chunks described by the TEMPLATE.  Each
7885           chunk is converted separately to a value.  Typically, either the
7886           string is a result of "pack", or the characters of the string
7887           represent a C structure of some kind.
7888
7889           The TEMPLATE has the same format as in the "pack" function.  Here's
7890           a subroutine that does substring:
7891
7892               sub substr {
7893                   my ($what, $where, $howmuch) = @_;
7894                   unpack("x$where a$howmuch", $what);
7895               }
7896
7897           and then there's
7898
7899               sub ordinal { unpack("W",$_[0]); } # same as ord()
7900
7901           In addition to fields allowed in "pack", you may prefix a field
7902           with a %<number> to indicate that you want a <number>-bit checksum
7903           of the items instead of the items themselves.  Default is a 16-bit
7904           checksum.  The checksum is calculated by summing numeric values of
7905           expanded values (for string fields the sum of "ord($char)" is
7906           taken; for bit fields the sum of zeroes and ones).
7907
7908           For example, the following computes the same number as the System V
7909           sum program:
7910
7911               my $checksum = do {
7912                   local $/;  # slurp!
7913                   unpack("%32W*", readline) % 65535;
7914               };
7915
7916           The following efficiently counts the number of set bits in a bit
7917           vector:
7918
7919               my $setbits = unpack("%32b*", $selectmask);
7920
7921           The "p" and "P" formats should be used with care.  Since Perl has
7922           no way of checking whether the value passed to "unpack" corresponds
7923           to a valid memory location, passing a pointer value that's not
7924           known to be valid is likely to have disastrous consequences.
7925
7926           If there are more pack codes or if the repeat count of a field or a
7927           group is larger than what the remainder of the input string allows,
7928           the result is not well defined: the repeat count may be decreased,
7929           or "unpack" may produce empty strings or zeros, or it may raise an
7930           exception.  If the input string is longer than one described by the
7931           TEMPLATE, the remainder of that input string is ignored.
7932
7933           See "pack" for more examples and notes.
7934
7935       unshift ARRAY,LIST
7936           Does the opposite of a "shift".  Or the opposite of a "push",
7937           depending on how you look at it.  Prepends list to the front of the
7938           array and returns the new number of elements in the array.
7939
7940               unshift(@ARGV, '-e') unless $ARGV[0] =~ /^-/;
7941
7942           Note the LIST is prepended whole, not one element at a time, so the
7943           prepended elements stay in the same order.  Use "reverse" to do the
7944           reverse.
7945
7946           Starting with Perl 5.14, an experimental feature allowed "unshift"
7947           to take a scalar expression. This experiment has been deemed
7948           unsuccessful, and was removed as of Perl 5.24.
7949
7950       untie VARIABLE
7951           Breaks the binding between a variable and a package.  (See tie.)
7952           Has no effect if the variable is not tied.
7953
7954       use Module VERSION LIST
7955       use Module VERSION
7956       use Module LIST
7957       use Module
7958           Imports some semantics into the current package from the named
7959           module, generally by aliasing certain subroutine or variable names
7960           into your package.  It is exactly equivalent to
7961
7962               BEGIN { require Module; Module->import( LIST ); }
7963
7964           except that Module must be a bareword.  The importation can be made
7965           conditional by using the if module.
7966
7967           The "BEGIN" forces the "require" and "import" to happen at compile
7968           time.  The "require" makes sure the module is loaded into memory if
7969           it hasn't been yet.  The "import" is not a builtin; it's just an
7970           ordinary static method call into the "Module" package to tell the
7971           module to import the list of features back into the current
7972           package.  The module can implement its "import" method any way it
7973           likes, though most modules just choose to derive their "import"
7974           method via inheritance from the "Exporter" class that is defined in
7975           the "Exporter" module.  See Exporter.  If no "import" method can be
7976           found, then the call is skipped, even if there is an AUTOLOAD
7977           method.
7978
7979           If you do not want to call the package's "import" method (for
7980           instance, to stop your namespace from being altered), explicitly
7981           supply the empty list:
7982
7983               use Module ();
7984
7985           That is exactly equivalent to
7986
7987               BEGIN { require Module }
7988
7989           If the VERSION argument is present between Module and LIST, then
7990           the "use" will call the "VERSION" method in class Module with the
7991           given version as an argument:
7992
7993               use Module 12.34;
7994
7995           is equivalent to:
7996
7997               BEGIN { require Module; Module->VERSION(12.34) }
7998
7999           The default "VERSION" method, inherited from the "UNIVERSAL" class,
8000           croaks if the given version is larger than the value of the
8001           variable $Module::VERSION.
8002
8003           The VERSION argument cannot be an arbitrary expression.  It only
8004           counts as a VERSION argument if it is a version number literal,
8005           starting with either a digit or "v" followed by a digit.  Anything
8006           that doesn't look like a version literal will be parsed as the
8007           start of the LIST.  Nevertheless, many attempts to use an arbitrary
8008           expression as a VERSION argument will appear to work, because
8009           Exporter's "import" method handles numeric arguments specially,
8010           performing version checks rather than treating them as things to
8011           export.
8012
8013           Again, there is a distinction between omitting LIST ("import"
8014           called with no arguments) and an explicit empty LIST "()" ("import"
8015           not called).  Note that there is no comma after VERSION!
8016
8017           Because this is a wide-open interface, pragmas (compiler
8018           directives) are also implemented this way.  Some of the currently
8019           implemented pragmas are:
8020
8021               use constant;
8022               use diagnostics;
8023               use integer;
8024               use sigtrap  qw(SEGV BUS);
8025               use strict   qw(subs vars refs);
8026               use subs     qw(afunc blurfl);
8027               use warnings qw(all);
8028               use sort     qw(stable);
8029
8030           Some of these pseudo-modules import semantics into the current
8031           block scope (like "strict" or "integer", unlike ordinary modules,
8032           which import symbols into the current package (which are effective
8033           through the end of the file).
8034
8035           Because "use" takes effect at compile time, it doesn't respect the
8036           ordinary flow control of the code being compiled.  In particular,
8037           putting a "use" inside the false branch of a conditional doesn't
8038           prevent it from being processed.  If a module or pragma only needs
8039           to be loaded conditionally, this can be done using the if pragma:
8040
8041               use if $] < 5.008, "utf8";
8042               use if WANT_WARNINGS, warnings => qw(all);
8043
8044           There's a corresponding "no" declaration that unimports meanings
8045           imported by "use", i.e., it calls "Module->unimport(LIST)" instead
8046           of "import".  It behaves just as "import" does with VERSION, an
8047           omitted or empty LIST, or no unimport method being found.
8048
8049               no integer;
8050               no strict 'refs';
8051               no warnings;
8052
8053           See perlmodlib for a list of standard modules and pragmas.  See
8054           perlrun for the "-M" and "-m" command-line options to Perl that
8055           give "use" functionality from the command-line.
8056
8057       use VERSION
8058           Lexically enables all features available in the requested version
8059           as defined by the feature pragma, disabling any features not in the
8060           requested version's feature bundle.  See feature.
8061
8062           VERSION may be either a v-string such as v5.24.1, which will be
8063           compared to $^V (aka $PERL_VERSION), or a numeric argument of the
8064           form 5.024001, which will be compared to $].  An exception is
8065           raised if VERSION is greater than the version of the current Perl
8066           interpreter; Perl will not attempt to parse the rest of the file.
8067           Compare with "require", which can do a similar check at run time.
8068
8069           If the specified Perl version is 5.12 or higher, strictures are
8070           enabled lexically as with "use strict".  Similarly, if the
8071           specified Perl version is 5.35.0 or higher, warnings are enabled.
8072           Later use of "use VERSION" will override all behavior of a previous
8073           "use VERSION", possibly removing the "strict", "warnings", and
8074           "feature" added by it.  "use VERSION" does not load the feature.pm,
8075           strict.pm, or warnings.pm files.
8076
8077           In the current implementation, any explicit use of "use strict" or
8078           "no strict" overrides "use VERSION", even if it comes before it.
8079           However, this may be subject to change in a future release of Perl,
8080           so new code should not rely on this fact.  It is recommended that a
8081           "use VERSION" declaration be the first significant statement within
8082           a file (possibly after a "package" statement or any amount of
8083           whitespace or comment), so that its effects happen first, and other
8084           pragmata are applied after it.
8085
8086           Specifying VERSION as a numeric argument of the form 5.024001
8087           should generally be avoided as older less readable syntax compared
8088           to v5.24.1. Before perl 5.8.0 released in 2002 the more verbose
8089           numeric form was the only supported syntax, which is why you might
8090           see it in older code.
8091
8092               use v5.24.1;    # compile time version check
8093               use 5.24.1;     # ditto
8094               use 5.024_001;  # ditto; older syntax compatible with perl 5.6
8095
8096           This is often useful if you need to check the current Perl version
8097           before "use"ing library modules that won't work with older versions
8098           of Perl.  (We try not to do this more than we have to.)
8099
8100           Symmetrically, "no VERSION" allows you to specify that you want a
8101           version of Perl older than the specified one.  Historically this
8102           was added during early designs of the Raku language (formerly "Perl
8103           6"), so that a Perl 5 program could begin
8104
8105               no 6;
8106
8107           to declare that it is not a Perl 6 program.  As the two languages
8108           have different implementations, file naming conventions, and other
8109           infrastructure, this feature is now little used in practice and
8110           should be avoided in newly-written code.
8111
8112           Care should be taken when using the "no VERSION" form, as it is
8113           only meant to be used to assert that the running Perl is of a
8114           earlier version than its argument and not to undo the feature-
8115           enabling side effects of "use VERSION".
8116
8117       utime LIST
8118           Changes the access and modification times on each file of a list of
8119           files.  The first two elements of the list must be the NUMERIC
8120           access and modification times, in that order.  Returns the number
8121           of files successfully changed.  The inode change time of each file
8122           is set to the current time.  For example, this code has the same
8123           effect as the Unix touch(1) command when the files already exist
8124           and belong to the user running the program:
8125
8126               #!/usr/bin/perl
8127               my $atime = my $mtime = time;
8128               utime $atime, $mtime, @ARGV;
8129
8130           Since Perl 5.8.0, if the first two elements of the list are
8131           "undef", the utime(2) syscall from your C library is called with a
8132           null second argument.  On most systems, this will set the file's
8133           access and modification times to the current time (i.e., equivalent
8134           to the example above) and will work even on files you don't own
8135           provided you have write permission:
8136
8137               for my $file (@ARGV) {
8138                   utime(undef, undef, $file)
8139                       || warn "Couldn't touch $file: $!";
8140               }
8141
8142           Under NFS this will use the time of the NFS server, not the time of
8143           the local machine.  If there is a time synchronization problem, the
8144           NFS server and local machine will have different times.  The Unix
8145           touch(1) command will in fact normally use this form instead of the
8146           one shown in the first example.
8147
8148           Passing only one of the first two elements as "undef" is equivalent
8149           to passing a 0 and will not have the effect described when both are
8150           "undef".  This also triggers an uninitialized warning.
8151
8152           On systems that support futimes(2), you may pass filehandles among
8153           the files.  On systems that don't support futimes(2), passing
8154           filehandles raises an exception.  Filehandles must be passed as
8155           globs or glob references to be recognized; barewords are considered
8156           filenames.
8157
8158           Portability issues: "utime" in perlport.
8159
8160       values HASH
8161       values ARRAY
8162           In list context, returns a list consisting of all the values of the
8163           named hash.  In Perl 5.12 or later only, will also return a list of
8164           the values of an array; prior to that release, attempting to use an
8165           array argument will produce a syntax error.  In scalar context,
8166           returns the number of values.
8167
8168           Hash entries are returned in an apparently random order.  The
8169           actual random order is specific to a given hash; the exact same
8170           series of operations on two hashes may result in a different order
8171           for each hash.  Any insertion into the hash may change the order,
8172           as will any deletion, with the exception that the most recent key
8173           returned by "each" or "keys" may be deleted without changing the
8174           order.  So long as a given hash is unmodified you may rely on
8175           "keys", "values" and "each" to repeatedly return the same order as
8176           each other.  See "Algorithmic Complexity Attacks" in perlsec for
8177           details on why hash order is randomized.  Aside from the guarantees
8178           provided here the exact details of Perl's hash algorithm and the
8179           hash traversal order are subject to change in any release of Perl.
8180           Tied hashes may behave differently to Perl's hashes with respect to
8181           changes in order on insertion and deletion of items.
8182
8183           As a side effect, calling "values" resets the HASH or ARRAY's
8184           internal iterator (see "each") before yielding the values.  In
8185           particular, calling "values" in void context resets the iterator
8186           with no other overhead.
8187
8188           Apart from resetting the iterator, "values @array" in list context
8189           is the same as plain @array.  (We recommend that you use void
8190           context "keys @array" for this, but reasoned that taking "values
8191           @array" out would require more documentation than leaving it in.)
8192
8193           Note that the values are not copied, which means modifying them
8194           will modify the contents of the hash:
8195
8196               for (values %hash)      { s/foo/bar/g }  # modifies %hash values
8197               for (@hash{keys %hash}) { s/foo/bar/g }  # same
8198
8199           Starting with Perl 5.14, an experimental feature allowed "values"
8200           to take a scalar expression. This experiment has been deemed
8201           unsuccessful, and was removed as of Perl 5.24.
8202
8203           To avoid confusing would-be users of your code who are running
8204           earlier versions of Perl with mysterious syntax errors, put this
8205           sort of thing at the top of your file to signal that your code will
8206           work only on Perls of a recent vintage:
8207
8208               use v5.12;  # so keys/values/each work on arrays
8209
8210           See also "keys", "each", and "sort".
8211
8212       vec EXPR,OFFSET,BITS
8213           Treats the string in EXPR as a bit vector made up of elements of
8214           width BITS and returns the value of the element specified by OFFSET
82