3 perlrun - how to execute the Perl interpreter
7 B<perl> S<[ B<-sTtuUWX> ]>
8 S<[ B<-hv> ] [ B<-V>[:I<configvar>] ]>
9 S<[ B<-cw> ] [ B<-d>[:I<debugger>] ] [ B<-D>[I<number/list>] ]>
10 S<[ B<-pna> ] [ B<-F>I<pattern> ] [ B<-l>[I<octal>] ] [ B<-0>[I<octal/hexadecimal>] ]>
11 S<[ B<-I>I<dir> ] [ B<-m>[B<->]I<module> ] [ B<-M>[B<->]I<'module...'> ]>
15 S<[ B<-i>[I<extension>] ]>
16 S<[ B<-e> I<'command'> ] [ B<--> ] [ I<programfile> ] [ I<argument> ]...>
17 S<[ B<-A [I<assertions>] >]>
18 S<[ B<-C [I<number/list>] >]>
22 The normal way to run a Perl program is by making it directly
23 executable, or else by passing the name of the source file as an
24 argument on the command line. (An interactive Perl environment
25 is also possible--see L<perldebug> for details on how to do that.)
26 Upon startup, Perl looks for your program in one of the following
33 Specified line by line via B<-e> switches on the command line.
37 Contained in the file specified by the first filename on the command line.
38 (Note that systems supporting the #! notation invoke interpreters this
39 way. See L<Location of Perl>.)
43 Passed in implicitly via standard input. This works only if there are
44 no filename arguments--to pass arguments to a STDIN-read program you
45 must explicitly specify a "-" for the program name.
49 With methods 2 and 3, Perl starts parsing the input file from the
50 beginning, unless you've specified a B<-x> switch, in which case it
51 scans for the first line starting with #! and containing the word
52 "perl", and starts there instead. This is useful for running a program
53 embedded in a larger message. (In this case you would indicate the end
54 of the program using the C<__END__> token.)
56 The #! line is always examined for switches as the line is being
57 parsed. Thus, if you're on a machine that allows only one argument
58 with the #! line, or worse, doesn't even recognize the #! line, you
59 still can get consistent switch behavior regardless of how Perl was
60 invoked, even if B<-x> was used to find the beginning of the program.
62 Because historically some operating systems silently chopped off
63 kernel interpretation of the #! line after 32 characters, some
64 switches may be passed in on the command line, and some may not;
65 you could even get a "-" without its letter, if you're not careful.
66 You probably want to make sure that all your switches fall either
67 before or after that 32-character boundary. Most switches don't
68 actually care if they're processed redundantly, but getting a "-"
69 instead of a complete switch could cause Perl to try to execute
70 standard input instead of your program. And a partial B<-I> switch
71 could also cause odd results.
73 Some switches do care if they are processed twice, for instance
74 combinations of B<-l> and B<-0>. Either put all the switches after
75 the 32-character boundary (if applicable), or replace the use of
76 B<-0>I<digits> by C<BEGIN{ $/ = "\0digits"; }>.
78 Parsing of the #! switches starts wherever "perl" is mentioned in the line.
79 The sequences "-*" and "- " are specifically ignored so that you could,
80 if you were so inclined, say
82 #!/bin/sh -- # -*- perl -*- -p
83 eval 'exec perl -wS $0 ${1+"$@"}'
84 if $running_under_some_shell;
86 to let Perl see the B<-p> switch.
88 A similar trick involves the B<env> program, if you have it.
92 The examples above use a relative path to the perl interpreter,
93 getting whatever version is first in the user's path. If you want
94 a specific version of Perl, say, perl5.005_57, you should place
95 that directly in the #! line's path.
97 If the #! line does not contain the word "perl", the program named after
98 the #! is executed instead of the Perl interpreter. This is slightly
99 bizarre, but it helps people on machines that don't do #!, because they
100 can tell a program that their SHELL is F</usr/bin/perl>, and Perl will then
101 dispatch the program to the correct interpreter for them.
103 After locating your program, Perl compiles the entire program to an
104 internal form. If there are any compilation errors, execution of the
105 program is not attempted. (This is unlike the typical shell script,
106 which might run part-way through before finding a syntax error.)
108 If the program is syntactically correct, it is executed. If the program
109 runs off the end without hitting an exit() or die() operator, an implicit
110 C<exit(0)> is provided to indicate successful completion.
112 =head2 #! and quoting on non-Unix systems
114 Unix's #! technique can be simulated on other systems:
122 extproc perl -S -your_switches
124 as the first line in C<*.cmd> file (B<-S> due to a bug in cmd.exe's
129 Create a batch file to run your program, and codify it in
130 C<ALTERNATE_SHEBANG> (see the F<dosish.h> file in the source
131 distribution for more information).
135 The Win95/NT installation, when using the ActiveState installer for Perl,
136 will modify the Registry to associate the F<.pl> extension with the perl
137 interpreter. If you install Perl by other means (including building from
138 the sources), you may have to modify the Registry yourself. Note that
139 this means you can no longer tell the difference between an executable
140 Perl program and a Perl library file.
144 A Macintosh perl program will have the appropriate Creator and
145 Type, so that double-clicking them will invoke the perl application.
151 $ perl -mysw 'f$env("procedure")' 'p1' 'p2' 'p3' 'p4' 'p5' 'p6' 'p7' 'p8' !
152 $ exit++ + ++$status != 0 and $exit = $status = undef;
154 at the top of your program, where B<-mysw> are any command line switches you
155 want to pass to Perl. You can now invoke the program directly, by saying
156 C<perl program>, or as a DCL procedure, by saying C<@program> (or implicitly
157 via F<DCL$PATH> by just using the name of the program).
159 This incantation is a bit much to remember, but Perl will display it for
160 you if you say C<perl "-V:startperl">.
164 Command-interpreters on non-Unix systems have rather different ideas
165 on quoting than Unix shells. You'll need to learn the special
166 characters in your command-interpreter (C<*>, C<\> and C<"> are
167 common) and how to protect whitespace and these characters to run
168 one-liners (see B<-e> below).
170 On some systems, you may have to change single-quotes to double ones,
171 which you must I<not> do on Unix or Plan 9 systems. You might also
172 have to change a single % to a %%.
177 perl -e 'print "Hello world\n"'
180 perl -e "print \"Hello world\n\""
183 print "Hello world\n"
184 (then Run "Myscript" or Shift-Command-R)
187 perl -e "print ""Hello world\n"""
189 The problem is that none of this is reliable: it depends on the
190 command and it is entirely possible neither works. If B<4DOS> were
191 the command shell, this would probably work better:
193 perl -e "print <Ctrl-x>"Hello world\n<Ctrl-x>""
195 B<CMD.EXE> in Windows NT slipped a lot of standard Unix functionality in
196 when nobody was looking, but just try to find documentation for its
199 Under the Macintosh, it depends which environment you are using. The MacPerl
200 shell, or MPW, is much like Unix shells in its support for several
201 quoting variants, except that it makes free use of the Macintosh's non-ASCII
202 characters as control characters.
204 There is no general solution to all of this. It's just a mess.
206 =head2 Location of Perl
208 It may seem obvious to say, but Perl is useful only when users can
209 easily find it. When possible, it's good for both F</usr/bin/perl>
210 and F</usr/local/bin/perl> to be symlinks to the actual binary. If
211 that can't be done, system administrators are strongly encouraged
212 to put (symlinks to) perl and its accompanying utilities into a
213 directory typically found along a user's PATH, or in some other
214 obvious and convenient place.
216 In this documentation, C<#!/usr/bin/perl> on the first line of the program
217 will stand in for whatever method works on your system. You are
218 advised to use a specific path if you care about a specific version.
220 #!/usr/local/bin/perl5.00554
222 or if you just want to be running at least version, place a statement
223 like this at the top of your program:
227 =head2 Command Switches
229 As with all standard commands, a single-character switch may be
230 clustered with the following switch, if any.
232 #!/usr/bin/perl -spi.orig # same as -s -p -i.orig
238 =item B<-0>[I<octal/hexadecimal>]
240 specifies the input record separator (C<$/>) as an octal or
241 hexadecimal number. If there are no digits, the null character is the
242 separator. Other switches may precede or follow the digits. For
243 example, if you have a version of B<find> which can print filenames
244 terminated by the null character, you can say this:
246 find . -name '*.orig' -print0 | perl -n0e unlink
248 The special value 00 will cause Perl to slurp files in paragraph mode.
249 The value 0777 will cause Perl to slurp files whole because there is no
250 legal byte with that value.
252 If you want to specify any Unicode character, use the hexadecimal
253 format: C<-0xHHH...>, where the C<H> are valid hexadecimal digits.
254 (This means that you cannot use the C<-x> with a directory name that
255 consists of hexadecimal digits.)
257 =item B<-A [I<assertions>]>
259 Activates the assertions given after the switch as a comma-separated
260 list of assertion names. If no assertion name is given, activates all
261 assertions. See L<assertions>.
265 turns on autosplit mode when used with a B<-n> or B<-p>. An implicit
266 split command to the @F array is done as the first thing inside the
267 implicit while loop produced by the B<-n> or B<-p>.
269 perl -ane 'print pop(@F), "\n";'
278 An alternate delimiter may be specified using B<-F>.
280 =item B<-C [I<number/list>]>
282 The C<-C> flag controls some Unicode of the Perl Unicode features.
284 As of 5.8.1, the C<-C> can be followed either by a number or a list
285 of option letters. The letters, their numeric values, and effects
286 are as follows; listing the letters is equal to summing the numbers.
288 I 1 STDIN is assumed to be in UTF-8
289 O 2 STDOUT will be in UTF-8
290 E 4 STDERR will be in UTF-8
292 i 8 UTF-8 is the default PerlIO layer for input streams
293 o 16 UTF-8 is the default PerlIO layer for output streams
295 A 32 the @ARGV elements are expected to be strings encoded
297 L 64 normally the "IOEioA" are unconditional,
298 the L makes them conditional on the locale environment
299 variables (the LC_ALL, LC_TYPE, and LANG, in the order
300 of decreasing precedence) -- if the variables indicate
301 UTF-8, then the selected "IOEioA" are in effect
303 For example, C<-COE> and C<-C6> will both turn on UTF-8-ness on both
304 STDOUT and STDERR. Repeating letters is just redundant, not cumulative
307 The C<io> options mean that any subsequent open() (or similar I/O
308 operations) will have the C<:utf8> PerlIO layer implicitly applied
309 to them, in other words, UTF-8 is expected from any input stream,
310 and UTF-8 is produced to any output stream. This is just the default,
311 with explicit layers in open() and with binmode() one can manipulate
314 C<-C> on its own (not followed by any number or option list), or the
315 empty string C<""> for the C<PERL_UNICODE> environment variable, has the
316 same effect as C<-CSDL>. In other words, the standard I/O handles and
317 the default C<open()> layer are UTF-8-fied B<but> only if the locale
318 environment variables indicate a UTF-8 locale. This behaviour follows
319 the I<implicit> (and problematic) UTF-8 behaviour of Perl 5.8.0.
321 You can use C<-C0> (or C<"0"> for C<PERL_UNICODE>) to explicitly
322 disable all the above Unicode features.
324 The read-only magic variable C<${^UNICODE}> reflects the numeric value
325 of this setting. This is variable is set during Perl startup and is
326 thereafter read-only. If you want runtime effects, use the three-arg
327 open() (see L<perlfunc/open>), the two-arg binmode() (see L<perlfunc/binmode>),
328 and the C<open> pragma (see L<open>).
330 (In Perls earlier than 5.8.1 the C<-C> switch was a Win32-only switch
331 that enabled the use of Unicode-aware "wide system call" Win32 APIs.
332 This feature was practically unused, however, and the command line
333 switch was therefore "recycled".)
337 causes Perl to check the syntax of the program and then exit without
338 executing it. Actually, it I<will> execute C<BEGIN>, C<CHECK>, and
339 C<use> blocks, because these are considered as occurring outside the
340 execution of your program. C<INIT> and C<END> blocks, however, will
345 runs the program under the Perl debugger. See L<perldebug>.
347 =item B<-d:>I<foo[=bar,baz]>
349 runs the program under the control of a debugging, profiling, or
350 tracing module installed as Devel::foo. E.g., B<-d:DProf> executes
351 the program using the Devel::DProf profiler. As with the B<-M>
352 flag, options may be passed to the Devel::foo package where they
353 will be received and interpreted by the Devel::foo::import routine.
354 The comma-separated list of options must follow a C<=> character.
357 =item B<-D>I<letters>
361 sets debugging flags. To watch how it executes your program, use
362 B<-Dtls>. (This works only if debugging is compiled into your
363 Perl.) Another nice value is B<-Dx>, which lists your compiled
364 syntax tree. And B<-Dr> displays compiled regular expressions;
365 the format of the output is explained in L<perldebguts>.
367 As an alternative, specify a number instead of list of letters (e.g.,
368 B<-D14> is equivalent to B<-Dtls>):
370 1 p Tokenizing and parsing
372 with v, displays all stacks
373 4 l Context (loop) stack processing
375 16 o Method and overloading resolution
376 32 c String/numeric conversions
377 64 P Print profiling info, preprocessor command for -P, source file input state
378 128 m Memory allocation
379 256 f Format processing
380 512 r Regular expression parsing and execution
381 1024 x Syntax tree dump
382 2048 u Tainting checks
383 4096 (Obsolete, previously used for LEAKTEST)
384 8192 H Hash dump -- usurps values()
385 16384 X Scratchpad allocation
387 65536 S Thread synchronization
389 262144 R Include reference counts of dumped variables (eg when using -Ds)
390 524288 J Do not s,t,P-debug (Jump over) opcodes within package DB
391 1048576 v Verbose: use in conjunction with other flags
392 2097152 C Copy On Write
393 4194304 A Consistency checks on internal structures
394 8388608 q quiet - currently only suppressed the "EXECUTING" message
396 All these flags require B<-DDEBUGGING> when you compile the Perl
397 executable (but see L<Devel::Peek>, L<re> which may change this).
398 See the F<INSTALL> file in the Perl source distribution
399 for how to do this. This flag is automatically set if you include B<-g>
400 option when C<Configure> asks you about optimizer/debugger flags.
402 If you're just trying to get a print out of each line of Perl code
403 as it executes, the way that C<sh -x> provides for shell scripts,
404 you can't use Perl's B<-D> switch. Instead do this
406 # If you have "env" utility
407 env=PERLDB_OPTS="NonStop=1 AutoTrace=1 frame=2" perl -dS program
409 # Bourne shell syntax
410 $ PERLDB_OPTS="NonStop=1 AutoTrace=1 frame=2" perl -dS program
413 % (setenv PERLDB_OPTS "NonStop=1 AutoTrace=1 frame=2"; perl -dS program)
415 See L<perldebug> for details and variations.
417 =item B<-e> I<commandline>
419 may be used to enter one line of program. If B<-e> is given, Perl
420 will not look for a filename in the argument list. Multiple B<-e>
421 commands may be given to build up a multi-line script. Make sure
422 to use semicolons where you would in a normal program.
424 =item B<-F>I<pattern>
426 specifies the pattern to split on if B<-a> is also in effect. The
427 pattern may be surrounded by C<//>, C<"">, or C<''>, otherwise it will be
428 put in single quotes.
432 prints a summary of the options.
434 =item B<-i>[I<extension>]
436 specifies that files processed by the C<E<lt>E<gt>> construct are to be
437 edited in-place. It does this by renaming the input file, opening the
438 output file by the original name, and selecting that output file as the
439 default for print() statements. The extension, if supplied, is used to
440 modify the name of the old file to make a backup copy, following these
443 If no extension is supplied, no backup is made and the current file is
446 If the extension doesn't contain a C<*>, then it is appended to the
447 end of the current filename as a suffix. If the extension does
448 contain one or more C<*> characters, then each C<*> is replaced
449 with the current filename. In Perl terms, you could think of this
452 ($backup = $extension) =~ s/\*/$file_name/g;
454 This allows you to add a prefix to the backup file, instead of (or in
455 addition to) a suffix:
457 $ perl -pi'orig_*' -e 's/bar/baz/' fileA # backup to 'orig_fileA'
459 Or even to place backup copies of the original files into another
460 directory (provided the directory already exists):
462 $ perl -pi'old/*.orig' -e 's/bar/baz/' fileA # backup to 'old/fileA.orig'
464 These sets of one-liners are equivalent:
466 $ perl -pi -e 's/bar/baz/' fileA # overwrite current file
467 $ perl -pi'*' -e 's/bar/baz/' fileA # overwrite current file
469 $ perl -pi'.orig' -e 's/bar/baz/' fileA # backup to 'fileA.orig'
470 $ perl -pi'*.orig' -e 's/bar/baz/' fileA # backup to 'fileA.orig'
472 From the shell, saying
474 $ perl -p -i.orig -e "s/foo/bar/; ... "
476 is the same as using the program:
478 #!/usr/bin/perl -pi.orig
481 which is equivalent to
484 $extension = '.orig';
486 if ($ARGV ne $oldargv) {
487 if ($extension !~ /\*/) {
488 $backup = $ARGV . $extension;
491 ($backup = $extension) =~ s/\*/$ARGV/g;
493 rename($ARGV, $backup);
494 open(ARGVOUT, ">$ARGV");
501 print; # this prints to original filename
505 except that the B<-i> form doesn't need to compare $ARGV to $oldargv to
506 know when the filename has changed. It does, however, use ARGVOUT for
507 the selected filehandle. Note that STDOUT is restored as the default
508 output filehandle after the loop.
510 As shown above, Perl creates the backup file whether or not any output
511 is actually changed. So this is just a fancy way to copy files:
513 $ perl -p -i'/some/file/path/*' -e 1 file1 file2 file3...
515 $ perl -p -i'.orig' -e 1 file1 file2 file3...
517 You can use C<eof> without parentheses to locate the end of each input
518 file, in case you want to append to each file, or reset line numbering
519 (see example in L<perlfunc/eof>).
521 If, for a given file, Perl is unable to create the backup file as
522 specified in the extension then it will skip that file and continue on
523 with the next one (if it exists).
525 For a discussion of issues surrounding file permissions and B<-i>,
526 see L<perlfaq5/Why does Perl let me delete read-only files? Why does -i clobber protected files? Isn't this a bug in Perl?>.
528 You cannot use B<-i> to create directories or to strip extensions from
531 Perl does not expand C<~> in filenames, which is good, since some
532 folks use it for their backup files:
534 $ perl -pi~ -e 's/foo/bar/' file1 file2 file3...
536 Finally, the B<-i> switch does not impede execution when no
537 files are given on the command line. In this case, no backup is made
538 (the original file cannot, of course, be determined) and processing
539 proceeds from STDIN to STDOUT as might be expected.
541 =item B<-I>I<directory>
543 Directories specified by B<-I> are prepended to the search path for
544 modules (C<@INC>), and also tells the C preprocessor where to search for
545 include files. The C preprocessor is invoked with B<-P>; by default it
546 searches /usr/include and /usr/lib/perl.
548 =item B<-l>[I<octnum>]
550 enables automatic line-ending processing. It has two separate
551 effects. First, it automatically chomps C<$/> (the input record
552 separator) when used with B<-n> or B<-p>. Second, it assigns C<$\>
553 (the output record separator) to have the value of I<octnum> so
554 that any print statements will have that separator added back on.
555 If I<octnum> is omitted, sets C<$\> to the current value of
556 C<$/>. For instance, to trim lines to 80 columns:
558 perl -lpe 'substr($_, 80) = ""'
560 Note that the assignment C<$\ = $/> is done when the switch is processed,
561 so the input record separator can be different than the output record
562 separator if the B<-l> switch is followed by a B<-0> switch:
564 gnufind / -print0 | perl -ln0e 'print "found $_" if -p'
566 This sets C<$\> to newline and then sets C<$/> to the null character.
568 =item B<-m>[B<->]I<module>
570 =item B<-M>[B<->]I<module>
572 =item B<-M>[B<->]I<'module ...'>
574 =item B<-[mM]>[B<->]I<module=arg[,arg]...>
576 B<-m>I<module> executes C<use> I<module> C<();> before executing your
579 B<-M>I<module> executes C<use> I<module> C<;> before executing your
580 program. You can use quotes to add extra code after the module name,
581 e.g., C<'-Mmodule qw(foo bar)'>.
583 If the first character after the B<-M> or B<-m> is a dash (C<->)
584 then the 'use' is replaced with 'no'.
586 A little builtin syntactic sugar means you can also say
587 B<-mmodule=foo,bar> or B<-Mmodule=foo,bar> as a shortcut for
588 C<'-Mmodule qw(foo bar)'>. This avoids the need to use quotes when
589 importing symbols. The actual code generated by B<-Mmodule=foo,bar> is
590 C<use module split(/,/,q{foo,bar})>. Note that the C<=> form
591 removes the distinction between B<-m> and B<-M>.
595 causes Perl to assume the following loop around your program, which
596 makes it iterate over filename arguments somewhat like B<sed -n> or
601 ... # your program goes here
604 Note that the lines are not printed by default. See B<-p> to have
605 lines printed. If a file named by an argument cannot be opened for
606 some reason, Perl warns you about it and moves on to the next file.
608 Here is an efficient way to delete all files older than a week:
610 find . -mtime +7 -print | perl -nle unlink
612 This is faster than using the B<-exec> switch of B<find> because you don't
613 have to start a process on every filename found. It does suffer from
614 the bug of mishandling newlines in pathnames, which you can fix if
615 you follow the example under B<-0>.
617 C<BEGIN> and C<END> blocks may be used to capture control before or after
618 the implicit program loop, just as in B<awk>.
622 causes Perl to assume the following loop around your program, which
623 makes it iterate over filename arguments somewhat like B<sed>:
628 ... # your program goes here
630 print or die "-p destination: $!\n";
633 If a file named by an argument cannot be opened for some reason, Perl
634 warns you about it, and moves on to the next file. Note that the
635 lines are printed automatically. An error occurring during printing is
636 treated as fatal. To suppress printing use the B<-n> switch. A B<-p>
637 overrides a B<-n> switch.
639 C<BEGIN> and C<END> blocks may be used to capture control before or after
640 the implicit loop, just as in B<awk>.
644 B<NOTE: Use of -P is strongly discouraged because of its inherent
645 problems, including poor portability.>
647 This option causes your program to be run through the C preprocessor before
648 compilation by Perl. Because both comments and B<cpp> directives begin
649 with the # character, you should avoid starting comments with any words
650 recognized by the C preprocessor such as C<"if">, C<"else">, or C<"define">.
652 If you're considering using C<-P>, you might also want to look at the
653 Filter::cpp module from CPAN.
655 The problems of -P include, but are not limited to:
661 The C<#!> line is stripped, so any switches there don't apply.
665 A C<-P> on a C<#!> line doesn't work.
669 B<All> lines that begin with (whitespace and) a C<#> but
670 do not look like cpp commands, are stripped, including anything
671 inside Perl strings, regular expressions, and here-docs .
675 In some platforms the C preprocessor knows too much: it knows about
676 the C++ -style until-end-of-line comments starting with C<"//">.
677 This will cause problems with common Perl constructs like
681 because after -P this will became illegal code
685 The workaround is to use some other quoting separator than C<"/">,
686 like for example C<"!">:
694 It requires not only a working C preprocessor but also a working
695 F<sed>. If not on UNIX, you are probably out of luck on this.
699 Script line numbers are not preserved.
703 The C<-x> does not work with C<-P>.
709 enables rudimentary switch parsing for switches on the command
710 line after the program name but before any filename arguments (or before
711 an argument of B<-->). This means you can have switches with two leading
712 dashes (B<--help>). Any switch found there is removed from @ARGV and sets the
713 corresponding variable in the Perl program. The following program
714 prints "1" if the program is invoked with a B<-xyz> switch, and "abc"
715 if it is invoked with B<-xyz=abc>.
718 if ($xyz) { print "$xyz\n" }
720 Do note that B<--help> creates the variable ${-help}, which is not compliant
725 makes Perl use the PATH environment variable to search for the
726 program (unless the name of the program contains directory separators).
728 On some platforms, this also makes Perl append suffixes to the
729 filename while searching for it. For example, on Win32 platforms,
730 the ".bat" and ".cmd" suffixes are appended if a lookup for the
731 original name fails, and if the name does not already end in one
732 of those suffixes. If your Perl was compiled with DEBUGGING turned
733 on, using the -Dp switch to Perl shows how the search progresses.
735 Typically this is used to emulate #! startup on platforms that
736 don't support #!. This example works on many platforms that
737 have a shell compatible with Bourne shell:
740 eval 'exec /usr/bin/perl -wS $0 ${1+"$@"}'
741 if $running_under_some_shell;
743 The system ignores the first line and feeds the program to F</bin/sh>,
744 which proceeds to try to execute the Perl program as a shell script.
745 The shell executes the second line as a normal shell command, and thus
746 starts up the Perl interpreter. On some systems $0 doesn't always
747 contain the full pathname, so the B<-S> tells Perl to search for the
748 program if necessary. After Perl locates the program, it parses the
749 lines and ignores them because the variable $running_under_some_shell
750 is never true. If the program will be interpreted by csh, you will need
751 to replace C<${1+"$@"}> with C<$*>, even though that doesn't understand
752 embedded spaces (and such) in the argument list. To start up sh rather
753 than csh, some systems may have to replace the #! line with a line
754 containing just a colon, which will be politely ignored by Perl. Other
755 systems can't control that, and need a totally devious construct that
756 will work under any of B<csh>, B<sh>, or Perl, such as the following:
758 eval '(exit $?0)' && eval 'exec perl -wS $0 ${1+"$@"}'
759 & eval 'exec /usr/bin/perl -wS $0 $argv:q'
760 if $running_under_some_shell;
762 If the filename supplied contains directory separators (i.e., is an
763 absolute or relative pathname), and if that file is not found,
764 platforms that append file extensions will do so and try to look
765 for the file with those extensions added, one by one.
767 On DOS-like platforms, if the program does not contain directory
768 separators, it will first be searched for in the current directory
769 before being searched for on the PATH. On Unix platforms, the
770 program will be searched for strictly on the PATH.
774 Like B<-T>, but taint checks will issue warnings rather than fatal
775 errors. These warnings can be controlled normally with C<no warnings
778 B<NOTE: this is not a substitute for -T.> This is meant only to be
779 used as a temporary development aid while securing legacy code:
780 for real production code and for new secure code written from scratch
781 always use the real B<-T>.
785 forces "taint" checks to be turned on so you can test them. Ordinarily
786 these checks are done only when running setuid or setgid. It's a
787 good idea to turn them on explicitly for programs that run on behalf
788 of someone else whom you might not necessarily trust, such as CGI
789 programs or any internet servers you might write in Perl. See
790 L<perlsec> for details. For security reasons, this option must be
791 seen by Perl quite early; usually this means it must appear early
792 on the command line or in the #! line for systems which support
797 This obsolete switch causes Perl to dump core after compiling your
798 program. You can then in theory take this core dump and turn it
799 into an executable file by using the B<undump> program (not supplied).
800 This speeds startup at the expense of some disk space (which you
801 can minimize by stripping the executable). (Still, a "hello world"
802 executable comes out to about 200K on my machine.) If you want to
803 execute a portion of your program before dumping, use the dump()
804 operator instead. Note: availability of B<undump> is platform
805 specific and may not be available for a specific port of Perl.
807 This switch has been superseded in favor of the new Perl code
808 generator backends to the compiler. See L<B> and L<B::Bytecode>
813 allows Perl to do unsafe operations. Currently the only "unsafe"
814 operations are the unlinking of directories while running as superuser,
815 and running setuid programs with fatal taint checks turned into
816 warnings. Note that the B<-w> switch (or the C<$^W> variable) must
817 be used along with this option to actually I<generate> the
818 taint-check warnings.
822 prints the version and patchlevel of your perl executable.
826 prints summary of the major perl configuration values and the current
831 Prints to STDOUT the value of the named configuration variable.
836 will provide strong clues about what your MANPATH variable should
837 be set to in order to access the Perl documentation.
841 prints warnings about dubious constructs, such as variable names
842 that are mentioned only once and scalar variables that are used
843 before being set, redefined subroutines, references to undefined
844 filehandles or filehandles opened read-only that you are attempting
845 to write on, values used as a number that doesn't look like numbers,
846 using an array as though it were a scalar, if your subroutines
847 recurse more than 100 deep, and innumerable other things.
849 This switch really just enables the internal C<$^W> variable. You
850 can disable or promote into fatal errors specific warnings using
851 C<__WARN__> hooks, as described in L<perlvar> and L<perlfunc/warn>.
852 See also L<perldiag> and L<perltrap>. A new, fine-grained warning
853 facility is also available if you want to manipulate entire classes
854 of warnings; see L<warnings> or L<perllexwarn>.
858 Enables all warnings regardless of C<no warnings> or C<$^W>.
863 Disables all warnings regardless of C<use warnings> or C<$^W>.
866 =item B<-x> I<directory>
868 tells Perl that the program is embedded in a larger chunk of unrelated
869 ASCII text, such as in a mail message. Leading garbage will be
870 discarded until the first line that starts with #! and contains the
871 string "perl". Any meaningful switches on that line will be applied.
872 If a directory name is specified, Perl will switch to that directory
873 before running the program. The B<-x> switch controls only the
874 disposal of leading garbage. The program must be terminated with
875 C<__END__> if there is trailing garbage to be ignored (the program
876 can process any or all of the trailing garbage via the DATA filehandle
887 Used if chdir has no argument.
891 Used if chdir has no argument and HOME is not set.
895 Used in executing subprocesses, and in finding the program if B<-S> is
900 A list of directories in which to look for Perl library
901 files before looking in the standard library and the current
902 directory. Any architecture-specific directories under the specified
903 locations are automatically included if they exist. If PERL5LIB is not
904 defined, PERLLIB is used. Directories are separated (like in PATH) by
905 a colon on unixish platforms and by a semicolon on Windows (the proper
906 path separator being given by the command C<perl -V:path_sep>).
908 When running taint checks (either because the program was running setuid
909 or setgid, or the B<-T> switch was used), neither variable is used.
910 The program should instead say:
912 use lib "/my/directory";
916 Command-line options (switches). Switches in this variable are taken
917 as if they were on every Perl command line. Only the B<-[DIMUdmtw]>
918 switches are allowed. When running taint checks (because the program
919 was running setuid or setgid, or the B<-T> switch was used), this
920 variable is ignored. If PERL5OPT begins with B<-T>, tainting will be
921 enabled, and any subsequent options ignored.
925 A space (or colon) separated list of PerlIO layers. If perl is built
926 to use PerlIO system for IO (the default) these layers effect perl's IO.
928 It is conventional to start layer names with a colon e.g. C<:perlio> to
929 emphasise their similarity to variable "attributes". But the code that parses
930 layer specification strings (which is also used to decode the PERLIO
931 environment variable) treats the colon as a separator.
933 An unset or empty PERLIO is equivalent to C<:stdio>.
935 The list becomes the default for I<all> perl's IO. Consequently only built-in
936 layers can appear in this list, as external layers (such as :encoding()) need
937 IO in order to load them!. See L<"open pragma"|open> for how to add external
938 encodings as defaults.
940 The layers that it makes sense to include in the PERLIO environment
941 variable are summarised below. For more details see L<PerlIO>.
947 A pseudolayer that turns I<off> the C<:utf8> flag for the layer below.
948 Unlikely to be useful on its own in the global PERLIO environment variable.
949 You perhaps were thinking of C<:crlf:bytes> or C<:perlio:bytes>.
953 A layer that implements DOS/Windows like CRLF line endings. On read
954 converts pairs of CR,LF to a single "\n" newline character. On write
955 converts each "\n" to a CR,LF pair. Note that this layer likes to be
956 one of its kind: it silently ignores attempts to be pushed into the
957 layer stack more than once.
959 (Gory details follow) To be more exact what happens is this: after
960 pushing itself to the stack, the C<:crlf> layer checks all the layers
961 below itself to find the first layer that is capable of being a CRLF
962 layer but is not yet enabled to be a CRLF layer. If it finds such a
963 layer, it enables the CRLFness of that other deeper layer, and then
964 pops itself off the stack. If not, fine, use the one we just pushed.
966 The end result is that a C<:crlf> means "please enable the first CRLF
967 layer you can find, and if you can't find one, here would be a good
968 spot to place a new one."
970 Based on the C<:perlio> layer.
974 A layer which implements "reading" of files by using C<mmap()> to
975 make (whole) file appear in the process's address space, and then
976 using that as PerlIO's "buffer". This I<may> be faster in certain
977 circumstances for large files, and may result in less physical memory
978 use when multiple processes are reading the same file.
980 Files which are not C<mmap()>-able revert to behaving like the C<:perlio>
981 layer. Writes also behave like C<:perlio> layer as C<mmap()> for write
982 needs extra house-keeping (to extend the file) which negates any advantage.
984 The C<:mmap> layer will not exist if platform does not support C<mmap()>.
988 A from scratch implementation of buffering for PerlIO. Provides fast
989 access to the buffer for C<sv_gets> which implements perl's readline/E<lt>E<gt>
990 and in general attempts to minimize data copying.
992 C<:perlio> will insert a C<:unix> layer below itself to do low level IO.
996 An experimental pseudolayer that removes the topmost layer.
997 Use with the same care as is reserved for nitroglyserin.
1001 A pseudolayer that manipulates other layers. Applying the <:raw>
1002 layer is equivalent to calling C<binmode($fh)>. It makes the stream
1003 pass each byte as-is without any translation. In particular CRLF
1004 translation, and/or :utf8 intuited from locale are disabled.
1006 Arranges for all accesses go straight to the lowest buffered layer provided
1007 by the configration. That is it strips off any layers above that layer.
1009 In Perl 5.6 and some books the C<:raw> layer (previously sometimes also
1010 referred to as a "discipline") is documented as the inverse of the
1011 C<:crlf> layer. That is no longer the case - other layers which would
1012 alter binary nature of the stream are also disabled. If you want UNIX
1013 line endings on a platform that normally does CRLF translation, but still
1014 want UTF-8 or encoding defaults the appropriate thing to do is to add
1015 C<:perlio> to PERLIO environment variable.
1019 This layer provides PerlIO interface by wrapping system's ANSI C "stdio"
1020 library calls. The layer provides both buffering and IO.
1021 Note that C<:stdio> layer does I<not> do CRLF translation even if that
1022 is platforms normal behaviour. You will need a C<:crlf> layer above it
1027 Lowest level layer which provides basic PerlIO operations in terms of
1028 UNIX/POSIX numeric file descriptor calls
1029 C<open(), read(), write(), lseek(), close()>
1033 A pseudolayer that turns on a flag on the layer below to tell perl
1034 that data sent to the stream should be converted to perl internal
1035 "utf8" form and that data from the stream should be considered as so
1036 encoded. On ASCII based platforms the encoding is UTF-8 and on EBCDIC
1037 platforms UTF-EBCDIC. May be useful in PERLIO environment variable to
1038 make UTF-8 the default. (To turn off that behaviour use C<:bytes>
1043 On Win32 platforms this I<experimental> layer uses native "handle" IO
1044 rather than unix-like numeric file descriptor layer. Known to be
1045 buggy in this release.
1049 On all platforms the default set of layers should give acceptable results.
1051 For UNIX platforms that will equivalent of "unix perlio" or "stdio".
1052 Configure is setup to prefer "stdio" implementation if system's library
1053 provides for fast access to the buffer, otherwise it uses the "unix perlio"
1056 On Win32 the default in this release is "unix crlf". Win32's "stdio"
1057 has a number of bugs/mis-features for perl IO which are somewhat
1058 C compiler vendor/version dependent. Using our own C<crlf> layer as
1059 the buffer avoids those issues and makes things more uniform.
1060 The C<crlf> layer provides CRLF to/from "\n" conversion as well as
1063 This release uses C<unix> as the bottom layer on Win32 and so still uses C
1064 compiler's numeric file descriptor routines. There is an experimental native
1065 C<win32> layer which is expected to be enhanced and should eventually replace
1070 If set to the name of a file or device then certain operations of PerlIO
1071 sub-system will be logged to that file (opened as append). Typical uses
1074 PERLIO_DEBUG=/dev/tty perl script ...
1076 and Win32 approximate equivalent:
1078 set PERLIO_DEBUG=CON
1084 A list of directories in which to look for Perl library
1085 files before looking in the standard library and the current directory.
1086 If PERL5LIB is defined, PERLLIB is not used.
1090 The command used to load the debugger code. The default is:
1092 BEGIN { require 'perl5db.pl' }
1094 =item PERL5SHELL (specific to the Win32 port)
1096 May be set to an alternative shell that perl must use internally for
1097 executing "backtick" commands or system(). Default is C<cmd.exe /x/d/c>
1098 on WindowsNT and C<command.com /c> on Windows95. The value is considered
1099 to be space-separated. Precede any character that needs to be protected
1100 (like a space or backslash) with a backslash.
1102 Note that Perl doesn't use COMSPEC for this purpose because
1103 COMSPEC has a high degree of variability among users, leading to
1104 portability concerns. Besides, perl can use a shell that may not be
1105 fit for interactive use, and setting COMSPEC to such a shell may
1106 interfere with the proper functioning of other programs (which usually
1107 look in COMSPEC to find a shell fit for interactive use).
1109 =item PERL_DEBUG_MSTATS
1111 Relevant only if perl is compiled with the malloc included with the perl
1112 distribution (that is, if C<perl -V:d_mymalloc> is 'define').
1113 If set, this causes memory statistics to be dumped after execution. If set
1114 to an integer greater than one, also causes memory statistics to be dumped
1117 =item PERL_DESTRUCT_LEVEL
1119 Relevant only if your perl executable was built with B<-DDEBUGGING>,
1120 this controls the behavior of global destruction of objects and other
1121 references. See L<perlhack/PERL_DESTRUCT_LEVEL> for more information.
1123 =item PERL_DL_NONLAZY
1125 Set to one to have perl resolve B<all> undefined symbols when it loads
1126 a dynamic library. The default behaviour is to resolve symbols when
1127 they are used. Setting this variable is useful during testing of
1128 extensions as it ensures that you get an error on misspelled function
1129 names even if the test suite doesn't call it.
1133 If using the C<encoding> pragma without an explicit encoding name, the
1134 PERL_ENCODING environment variable is consulted for an encoding name.
1136 =item PERL_HASH_SEED
1138 (Since Perl 5.8.1.) Used to randomise Perl's internal hash function.
1139 To emulate the pre-5.8.1 behaviour, set to an integer (zero means
1140 exactly the same order as 5.8.0). "Pre-5.8.1" means, among other
1141 things, that hash keys will be ordered the same between different runs
1144 The default behaviour is to randomise unless the PERL_HASH_SEED is set.
1145 If Perl has been compiled with C<-DUSE_HASH_SEED_EXPLICIT>, the default
1146 behaviour is B<not> to randomise unless the PERL_HASH_SEED is set.
1148 If PERL_HASH_SEED is unset or set to a non-numeric string, Perl uses
1149 the pseudorandom seed supplied by the operating system and libraries.
1150 This means that each different run of Perl will have a different
1151 ordering of the results of keys(), values(), and each().
1153 B<Please note that the hash seed is sensitive information>. Hashes are
1154 randomized to protect against local and remote attacks against Perl
1155 code. By manually setting a seed this protection may be partially or
1158 See L<perlsec/"Algorithmic Complexity Attacks"> and
1159 L</PERL_HASH_SEED_DEBUG> for more information.
1161 =item PERL_HASH_SEED_DEBUG
1163 (Since Perl 5.8.1.) Set to one to display (to STDERR) the value of
1164 the hash seed at the beginning of execution. This, combined with
1165 L</PERL_HASH_SEED> is intended to aid in debugging nondeterministic
1166 behavior caused by hash randomization.
1168 B<Note that the hash seed is sensitive information>: by knowing it one
1169 can craft a denial-of-service attack against Perl code, even remotely,
1170 see L<perlsec/"Algorithmic Complexity Attacks"> for more information.
1171 B<Do not disclose the hash seed> to people who don't need to know it.
1172 See also hash_seed() of L<Hash::Util>.
1174 =item PERL_ROOT (specific to the VMS port)
1176 A translation concealed rooted logical name that contains perl and the
1177 logical device for the @INC path on VMS only. Other logical names that
1178 affect perl on VMS include PERLSHR, PERL_ENV_TABLES, and
1179 SYS$TIMEZONE_DIFFERENTIAL but are optional and discussed further in
1180 L<perlvms> and in F<README.vms> in the Perl source distribution.
1184 In Perls 5.8.1 and later. If set to C<unsafe> the pre-Perl-5.8.0
1185 signals behaviour (immediate but unsafe) is restored. If set to
1186 C<safe> the safe (or deferred) signals are used.
1187 See L<perlipc/"Deferred Signals (Safe signals)">.
1191 Equivalent to the B<-C> command-line switch. Note that this is not
1192 a boolean variable-- setting this to C<"1"> is not the right way to
1193 "enable Unicode" (whatever that would mean). You can use C<"0"> to
1194 "disable Unicode", though (or alternatively unset PERL_UNICODE in
1195 your shell before starting Perl). See the description of the C<-C>
1196 switch for more information.
1198 =item SYS$LOGIN (specific to the VMS port)
1200 Used if chdir has no argument and HOME and LOGDIR are not set.
1204 Perl also has environment variables that control how Perl handles data
1205 specific to particular natural languages. See L<perllocale>.
1207 Apart from these, Perl uses no other environment variables, except
1208 to make them available to the program being executed, and to child
1209 processes. However, programs running setuid would do well to execute
1210 the following lines before doing anything else, just to keep people
1213 $ENV{PATH} = '/bin:/usr/bin'; # or whatever you need
1214 $ENV{SHELL} = '/bin/sh' if exists $ENV{SHELL};
1215 delete @ENV{qw(IFS CDPATH ENV BASH_ENV)};