1PERLRUN(1) Perl Programmers Reference Guide PERLRUN(1)
2
6 perlrun - how to execute the Perl interpreter
7
9 perl [ -sTtuUWX ] [ -hv ] [ -V[:configvar] ]
10 [ -cw ] [ -d[t][:debugger] ] [ -D[number/list] ]
11 [ -pna ] [ -Fpattern ] [ -l[octal] ] [ -0[octal/hexadecimal] ]
12 [ -Idir ] [ -m[-]module ] [ -M[-]'module...' ] [ -f ]
13 [ -C [number/list] ] [ -S ] [ -x[dir] ]
14 [ -i[extension] ]
15 [ [-e|-E] 'command' ] [ -- ] [ programfile ] [ argument ]...
16
18 The normal way to run a Perl program is by making it directly
19 executable, or else by passing the name of the source file as an
20 argument on the command line. (An interactive Perl environment is also
21 possible--see perldebug for details on how to do that.) Upon startup,
22 Perl looks for your program in one of the following places:
23 1. Specified line by line via -e or -E switches on the command line.
25 2. Contained in the file specified by the first filename on the
27 command line. (Note that systems supporting the "#!" notation
28 invoke interpreters this way. See "Location of Perl".)
29 3. Passed in implicitly via standard input. This works only if there
31 are no filename arguments--to pass arguments to a STDIN-read
32 program you must explicitly specify a "-" for the program name.
33 With methods 2 and 3, Perl starts parsing the input file from the
35 beginning, unless you've specified a "-x" switch, in which case it
36 scans for the first line starting with "#!" and containing the word
37 "perl", and starts there instead. This is useful for running a program
38 embedded in a larger message. (In this case you would indicate the end
39 of the program using the "__END__" token.)
40 The "#!" line is always examined for switches as the line is being
42 parsed. Thus, if you're on a machine that allows only one argument
43 with the "#!" line, or worse, doesn't even recognize the "#!" line, you
44 still can get consistent switch behaviour regardless of how Perl was
45 invoked, even if "-x" was used to find the beginning of the program.
46 Because historically some operating systems silently chopped off kernel
48 interpretation of the "#!" line after 32 characters, some switches may
49 be passed in on the command line, and some may not; you could even get
50 a "-" without its letter, if you're not careful. You probably want to
51 make sure that all your switches fall either before or after that
52 32-character boundary. Most switches don't actually care if they're
53 processed redundantly, but getting a "-" instead of a complete switch
54 could cause Perl to try to execute standard input instead of your
55 program. And a partial -I switch could also cause odd results.
56 Some switches do care if they are processed twice, for instance
58 combinations of -l and -0. Either put all the switches after the
59 32-character boundary (if applicable), or replace the use of -0digits
60 by "BEGIN{ $/ = "\0digits"; }".
61 Parsing of the "#!" switches starts wherever "perl" is mentioned in the
63 line. The sequences "-*" and "- " are specifically ignored so that you
64 could, if you were so inclined, say
65 #!/bin/sh
67 #! -*- perl -*- -p
68 eval 'exec perl -x -wS $0 ${1+"$@"}'
69 if 0;
70 to let Perl see the "-p" switch.
72 A similar trick involves the env program, if you have it.
74 #!/usr/bin/env perl
76 The examples above use a relative path to the perl interpreter, getting
78 whatever version is first in the user's path. If you want a specific
79 version of Perl, say, perl5.14.1, you should place that directly in the
80 "#!" line's path.
81 If the "#!" line does not contain the word "perl" nor the word "indir",
83 the program named after the "#!" is executed instead of the Perl
84 interpreter. This is slightly bizarre, but it helps people on machines
85 that don't do "#!", because they can tell a program that their SHELL is
86 /usr/bin/perl, and Perl will then dispatch the program to the correct
87 interpreter for them.
88 After locating your program, Perl compiles the entire program to an
90 internal form. If there are any compilation errors, execution of the
91 program is not attempted. (This is unlike the typical shell script,
92 which might run part-way through before finding a syntax error.)
93 If the program is syntactically correct, it is executed. If the
95 program runs off the end without hitting an exit() or die() operator,
96 an implicit exit(0) is provided to indicate successful completion.
97 #! and quoting on non-Unix systems
99 Unix's "#!" technique can be simulated on other systems:
100 OS/2
102 Put
103 extproc perl -S -your_switches
105 as the first line in "*.cmd" file ("-S" due to a bug in cmd.exe's
107 `extproc' handling).
108 MS-DOS
110 Create a batch file to run your program, and codify it in
111 "ALTERNATE_SHEBANG" (see the dosish.h file in the source
112 distribution for more information).
113 Win95/NT
115 The Win95/NT installation, when using the ActiveState installer for
116 Perl, will modify the Registry to associate the .pl extension with
117 the perl interpreter. If you install Perl by other means
118 (including building from the sources), you may have to modify the
119 Registry yourself. Note that this means you can no longer tell the
120 difference between an executable Perl program and a Perl library
121 file.
122 VMS Put
124 $ perl -mysw 'f$env("procedure")' 'p1' 'p2' 'p3' 'p4' 'p5' 'p6' 'p7' 'p8' !
126 $ exit++ + ++$status != 0 and $exit = $status = undef;
127 at the top of your program, where -mysw are any command line
129 switches you want to pass to Perl. You can now invoke the program
130 directly, by saying "perl program", or as a DCL procedure, by
131 saying @program (or implicitly via DCL$PATH by just using the name
132 of the program).
133 This incantation is a bit much to remember, but Perl will display
135 it for you if you say "perl "-V:startperl"".
136 Command-interpreters on non-Unix systems have rather different ideas on
138 quoting than Unix shells. You'll need to learn the special characters
139 in your command-interpreter ("*", "\" and """ are common) and how to
140 protect whitespace and these characters to run one-liners (see -e
141 below).
142 On some systems, you may have to change single-quotes to double ones,
144 which you must not do on Unix or Plan 9 systems. You might also have
145 to change a single % to a %%.
146 For example:
148 # Unix
150 perl -e 'print "Hello world\n"'
151 # MS-DOS, etc.
153 perl -e "print \"Hello world\n\""
154 # VMS
156 perl -e "print ""Hello world\n"""
157 The problem is that none of this is reliable: it depends on the command
159 and it is entirely possible neither works. If 4DOS were the command
160 shell, this would probably work better:
161 perl -e "print <Ctrl-x>"Hello world\n<Ctrl-x>""
163 CMD.EXE in Windows NT slipped a lot of standard Unix functionality in
165 when nobody was looking, but just try to find documentation for its
166 quoting rules.
167 There is no general solution to all of this. It's just a mess.
169 Location of Perl
171 It may seem obvious to say, but Perl is useful only when users can
172 easily find it. When possible, it's good for both /usr/bin/perl and
173 /usr/local/bin/perl to be symlinks to the actual binary. If that can't
174 be done, system administrators are strongly encouraged to put (symlinks
175 to) perl and its accompanying utilities into a directory typically
176 found along a user's PATH, or in some other obvious and convenient
177 place.
178 In this documentation, "#!/usr/bin/perl" on the first line of the
180 program will stand in for whatever method works on your system. You
181 are advised to use a specific path if you care about a specific
182 version.
183 #!/usr/local/bin/perl5.14
185 or if you just want to be running at least version, place a statement
187 like this at the top of your program:
188 use 5.014;
190 Command Switches
192 As with all standard commands, a single-character switch may be
193 clustered with the following switch, if any.
194 #!/usr/bin/perl -spi.orig # same as -s -p -i.orig
196 A "--" signals the end of options and disables further option
198 processing. Any arguments after the "--" are treated as filenames and
199 arguments.
200 Switches include:
202 -0[octal/hexadecimal]
204 specifies the input record separator ($/) as an octal or
205 hexadecimal number. If there are no digits, the null character is
206 the separator. Other switches may precede or follow the digits.
207 For example, if you have a version of find which can print
208 filenames terminated by the null character, you can say this:
209 find . -name '*.orig' -print0 | perl -n0e unlink
211 The special value 00 will cause Perl to slurp files in paragraph
213 mode. Any value 0400 or above will cause Perl to slurp files
214 whole, but by convention the value 0777 is the one normally used
215 for this purpose.
216 You can also specify the separator character using hexadecimal
218 notation: -0xHHH..., where the "H" are valid hexadecimal digits.
219 Unlike the octal form, this one may be used to specify any Unicode
220 character, even those beyond 0xFF. So if you really want a record
221 separator of 0777, specify it as -0x1FF. (This means that you
222 cannot use the "-x" option with a directory name that consists of
223 hexadecimal digits, or else Perl will think you have specified a
224 hex number to -0.)
225 -a turns on autosplit mode when used with a "-n" or "-p". An
227 implicit split command to the @F array is done as the first thing
228 inside the implicit while loop produced by the "-n" or "-p".
229 perl -ane 'print pop(@F), "\n";'
231 is equivalent to
233 while (<>) {
235 @F = split(' ');
236 print pop(@F), "\n";
237 }
238 An alternate delimiter may be specified using -F.
240 -a implicitly sets "-n".
242 -C [number/list]
244 The -C flag controls some of the Perl Unicode features.
245 As of 5.8.1, the -C can be followed either by a number or a list
247 of option letters. The letters, their numeric values, and effects
248 are as follows; listing the letters is equal to summing the
249 numbers.
250 I 1 STDIN is assumed to be in UTF-8
252 O 2 STDOUT will be in UTF-8
253 E 4 STDERR will be in UTF-8
254 S 7 I + O + E
255 i 8 UTF-8 is the default PerlIO layer for input streams
256 o 16 UTF-8 is the default PerlIO layer for output streams
257 D 24 i + o
258 A 32 the @ARGV elements are expected to be strings encoded
259 in UTF-8
260 L 64 normally the "IOEioA" are unconditional, the L makes
261 them conditional on the locale environment variables
262 (the LC_ALL, LC_CTYPE, and LANG, in the order of
263 decreasing precedence) -- if the variables indicate
264 UTF-8, then the selected "IOEioA" are in effect
265 a 256 Set ${^UTF8CACHE} to -1, to run the UTF-8 caching
266 code in debugging mode.
267 For example, -COE and -C6 will both turn on UTF-8-ness on both
269 STDOUT and STDERR. Repeating letters is just redundant, not
270 cumulative nor toggling.
271 The "io" options mean that any subsequent open() (or similar I/O
273 operations) in main program scope will have the ":utf8" PerlIO
274 layer implicitly applied to them, in other words, UTF-8 is
275 expected from any input stream, and UTF-8 is produced to any
276 output stream. This is just the default set via "${^OPEN}", with
277 explicit layers in open() and with binmode() one can manipulate
278 streams as usual. This has no effect on code run in modules.
279 -C on its own (not followed by any number or option list), or the
281 empty string "" for the "PERL_UNICODE" environment variable, has
282 the same effect as -CSDL. In other words, the standard I/O
283 handles and the default "open()" layer are UTF-8-fied but only if
284 the locale environment variables indicate a UTF-8 locale. This
285 behaviour follows the implicit (and problematic) UTF-8 behaviour
286 of Perl 5.8.0. (See "UTF-8 no longer default under UTF-8 locales"
287 in perl581delta.)
288 You can use -C0 (or "0" for "PERL_UNICODE") to explicitly disable
290 all the above Unicode features.
291 The read-only magic variable "${^UNICODE}" reflects the numeric
293 value of this setting. This variable is set during Perl startup
294 and is thereafter read-only. If you want runtime effects, use the
295 three-arg open() (see "open" in perlfunc), the two-arg binmode()
296 (see "binmode" in perlfunc), and the "open" pragma (see open).
297 (In Perls earlier than 5.8.1 the -C switch was a Win32-only switch
299 that enabled the use of Unicode-aware "wide system call" Win32
300 APIs. This feature was practically unused, however, and the
301 command line switch was therefore "recycled".)
302 Note: Since perl 5.10.1, if the -C option is used on the "#!"
304 line, it must be specified on the command line as well, since the
305 standard streams are already set up at this point in the execution
306 of the perl interpreter. You can also use binmode() to set the
307 encoding of an I/O stream.
308 -c causes Perl to check the syntax of the program and then exit
310 without executing it. Actually, it will execute any "BEGIN",
311 "UNITCHECK", or "CHECK" blocks and any "use" statements: these are
312 considered as occurring outside the execution of your program.
313 "INIT" and "END" blocks, however, will be skipped.
314 -d
316 -dt runs the program under the Perl debugger. See perldebug. If t is
317 specified, it indicates to the debugger that threads will be used
318 in the code being debugged.
319 -d:MOD[=bar,baz]
321 -dt:MOD[=bar,baz]
322 runs the program under the control of a debugging, profiling, or
323 tracing module installed as "Devel::MOD". E.g., -d:DProf executes
324 the program using the "Devel::DProf" profiler. As with the -M
325 flag, options may be passed to the "Devel::MOD" package where they
326 will be received and interpreted by the "Devel::MOD::import"
327 routine. Again, like -M, use --d:-MOD to call
328 "Devel::MOD::unimport" instead of import. The comma-separated
329 list of options must follow a "=" character. If t is specified,
330 it indicates to the debugger that threads will be used in the code
331 being debugged. See perldebug.
332 -Dletters
334 -Dnumber
335 sets debugging flags. This switch is enabled only if your perl
336 binary has been built with debugging enabled: normal production
337 perls won't have been.
338 For example, to watch how perl executes your program, use -Dtls.
340 Another nice value is -Dx, which lists your compiled syntax tree,
341 and -Dr displays compiled regular expressions; the format of the
342 output is explained in perldebguts.
343 As an alternative, specify a number instead of list of letters
345 (e.g., -D14 is equivalent to -Dtls):
346 1 p Tokenizing and parsing (with v, displays parse
348 stack)
349 2 s Stack snapshots (with v, displays all stacks)
350 4 l Context (loop) stack processing
351 8 t Trace execution
352 16 o Method and overloading resolution
353 32 c String/numeric conversions
354 64 P Print profiling info, source file input state
355 128 m Memory and SV allocation
356 256 f Format processing
357 512 r Regular expression parsing and execution
358 1024 x Syntax tree dump
359 2048 u Tainting checks
360 4096 U Unofficial, User hacking (reserved for private,
361 unreleased use)
362 16384 X Scratchpad allocation
363 32768 D Cleaning up
364 65536 S Op slab allocation
365 131072 T Tokenizing
366 262144 R Include reference counts of dumped variables
367 (eg when using -Ds)
368 524288 J show s,t,P-debug (don't Jump over) on opcodes within
369 package DB
370 1048576 v Verbose: use in conjunction with other flags to
371 increase the verbosity of the output. Is a no-op on
372 many of the other flags
373 2097152 C Copy On Write
374 4194304 A Consistency checks on internal structures
375 8388608 q quiet - currently only suppresses the "EXECUTING"
376 message
377 16777216 M trace smart match resolution
378 33554432 B dump suBroutine definitions, including special
379 Blocks like BEGIN
380 67108864 L trace Locale-related info; what gets output is very
381 subject to change
382 134217728 i trace PerlIO layer processing. Set PERLIO_DEBUG to
383 the filename to trace to.
384 268435456 y trace y///, tr/// compilation and execution
385 All these flags require -DDEBUGGING when you compile the Perl
387 executable (but see ":opd" in Devel::Peek or "'debug' mode" in re
388 which may change this). See the INSTALL file in the Perl source
389 distribution for how to do this.
390 If you're just trying to get a print out of each line of Perl code
392 as it executes, the way that "sh -x" provides for shell scripts,
393 you can't use Perl's -D switch. Instead do this
394 # If you have "env" utility
396 env PERLDB_OPTS="NonStop=1 AutoTrace=1 frame=2" perl -dS program
397 # Bourne shell syntax
399 $ PERLDB_OPTS="NonStop=1 AutoTrace=1 frame=2" perl -dS program
400 # csh syntax
402 % (setenv PERLDB_OPTS "NonStop=1 AutoTrace=1 frame=2"; perl -dS program)
403 See perldebug for details and variations.
405 -e commandline
407 may be used to enter one line of program. If -e is given, Perl
408 will not look for a filename in the argument list. Multiple -e
409 commands may be given to build up a multi-line script. Make sure
410 to use semicolons where you would in a normal program.
411 -E commandline
413 behaves just like -e, except that it implicitly enables all
414 optional features (in the main compilation unit). See feature.
415 -f Disable executing $Config{sitelib}/sitecustomize.pl at startup.
417 Perl can be built so that it by default will try to execute
419 $Config{sitelib}/sitecustomize.pl at startup (in a BEGIN block).
420 This is a hook that allows the sysadmin to customize how Perl
421 behaves. It can for instance be used to add entries to the @INC
422 array to make Perl find modules in non-standard locations.
423 Perl actually inserts the following code:
425 BEGIN {
427 do { local $!; -f "$Config{sitelib}/sitecustomize.pl"; }
428 && do "$Config{sitelib}/sitecustomize.pl";
429 }
430 Since it is an actual "do" (not a "require"), sitecustomize.pl
432 doesn't need to return a true value. The code is run in package
433 "main", in its own lexical scope. However, if the script dies, $@
434 will not be set.
435 The value of $Config{sitelib} is also determined in C code and not
437 read from "Config.pm", which is not loaded.
438 The code is executed very early. For example, any changes made to
440 @INC will show up in the output of `perl -V`. Of course, "END"
441 blocks will be likewise executed very late.
442 To determine at runtime if this capability has been compiled in
444 your perl, you can check the value of $Config{usesitecustomize}.
445 -Fpattern
447 specifies the pattern to split on for "-a". The pattern may be
448 surrounded by "//", "", or '', otherwise it will be put in single
449 quotes. You can't use literal whitespace or NUL characters in the
450 pattern.
451 -F implicitly sets both "-a" and "-n".
453 -h prints a summary of the options.
455 -i[extension]
457 specifies that files processed by the "<>" construct are to be
458 edited in-place. It does this by renaming the input file, opening
459 the output file by the original name, and selecting that output
460 file as the default for print() statements. The extension, if
461 supplied, is used to modify the name of the old file to make a
462 backup copy, following these rules:
463 If no extension is supplied, and your system supports it, the
465 original file is kept open without a name while the output is
466 redirected to a new file with the original filename. When perl
467 exits, cleanly or not, the original file is unlinked.
468 If the extension doesn't contain a "*", then it is appended to the
470 end of the current filename as a suffix. If the extension does
471 contain one or more "*" characters, then each "*" is replaced with
472 the current filename. In Perl terms, you could think of this as:
473 ($backup = $extension) =~ s/\*/$file_name/g;
475 This allows you to add a prefix to the backup file, instead of (or
477 in addition to) a suffix:
478 $ perl -pi'orig_*' -e 's/bar/baz/' fileA # backup to
480 # 'orig_fileA'
481 Or even to place backup copies of the original files into another
483 directory (provided the directory already exists):
484 $ perl -pi'old/*.orig' -e 's/bar/baz/' fileA # backup to
486 # 'old/fileA.orig'
487 These sets of one-liners are equivalent:
489 $ perl -pi -e 's/bar/baz/' fileA # overwrite current file
491 $ perl -pi'*' -e 's/bar/baz/' fileA # overwrite current file
492 $ perl -pi'.orig' -e 's/bar/baz/' fileA # backup to 'fileA.orig'
494 $ perl -pi'*.orig' -e 's/bar/baz/' fileA # backup to 'fileA.orig'
495 From the shell, saying
497 $ perl -p -i.orig -e "s/foo/bar/; ... "
499 is the same as using the program:
501 #!/usr/bin/perl -pi.orig
503 s/foo/bar/;
504 which is equivalent to
506 #!/usr/bin/perl
508 $extension = '.orig';
509 LINE: while (<>) {
510 if ($ARGV ne $oldargv) {
511 if ($extension !~ /\*/) {
512 $backup = $ARGV . $extension;
513 }
514 else {
515 ($backup = $extension) =~ s/\*/$ARGV/g;
516 }
517 rename($ARGV, $backup);
518 open(ARGVOUT, ">$ARGV");
519 select(ARGVOUT);
520 $oldargv = $ARGV;
521 }
522 s/foo/bar/;
523 }
524 continue {
525 print; # this prints to original filename
526 }
527 select(STDOUT);
528 except that the -i form doesn't need to compare $ARGV to $oldargv
530 to know when the filename has changed. It does, however, use
531 ARGVOUT for the selected filehandle. Note that STDOUT is restored
532 as the default output filehandle after the loop.
533 As shown above, Perl creates the backup file whether or not any
535 output is actually changed. So this is just a fancy way to copy
536 files:
537 $ perl -p -i'/some/file/path/*' -e 1 file1 file2 file3...
539 or
540 $ perl -p -i'.orig' -e 1 file1 file2 file3...
541 You can use "eof" without parentheses to locate the end of each
543 input file, in case you want to append to each file, or reset line
544 numbering (see example in "eof" in perlfunc).
545 If, for a given file, Perl is unable to create the backup file as
547 specified in the extension then it will skip that file and
548 continue on with the next one (if it exists).
549 For a discussion of issues surrounding file permissions and -i,
551 see "Why does Perl let me delete read-only files? Why does -i
552 clobber protected files? Isn't this a bug in Perl?" in perlfaq5.
553 You cannot use -i to create directories or to strip extensions
555 from files.
556 Perl does not expand "~" in filenames, which is good, since some
558 folks use it for their backup files:
559 $ perl -pi~ -e 's/foo/bar/' file1 file2 file3...
561 Note that because -i renames or deletes the original file before
563 creating a new file of the same name, Unix-style soft and hard
564 links will not be preserved.
565 Finally, the -i switch does not impede execution when no files are
567 given on the command line. In this case, no backup is made (the
568 original file cannot, of course, be determined) and processing
569 proceeds from STDIN to STDOUT as might be expected.
570 -Idirectory
572 Directories specified by -I are prepended to the search path for
573 modules (@INC).
574 -l[octnum]
576 enables automatic line-ending processing. It has two separate
577 effects. First, it automatically chomps $/ (the input record
578 separator) when used with "-n" or "-p". Second, it assigns "$\"
579 (the output record separator) to have the value of octnum so that
580 any print statements will have that separator added back on. If
581 octnum is omitted, sets "$\" to the current value of $/. For
582 instance, to trim lines to 80 columns:
583 perl -lpe 'substr($_, 80) = ""'
585 Note that the assignment "$\ = $/" is done when the switch is
587 processed, so the input record separator can be different than the
588 output record separator if the -l switch is followed by a -0
589 switch:
590 gnufind / -print0 | perl -ln0e 'print "found $_" if -p'
592 This sets "$\" to newline and then sets $/ to the null character.
594 -m[-]module
596 -M[-]module
597 -M[-]'module ...'
598 -[mM][-]module=arg[,arg]...
599 -mmodule executes "use" module "();" before executing your
600 program. This loads the module, but does not call its "import"
601 method, so does not import subroutines and does not give effect to
602 a pragma.
603 -Mmodule executes "use" module ";" before executing your program.
605 This loads the module and calls its "import" method, causing the
606 module to have its default effect, typically importing subroutines
607 or giving effect to a pragma. You can use quotes to add extra
608 code after the module name, e.g., '-MMODULE qw(foo bar)'.
609 If the first character after the -M or -m is a dash (-) then the
611 'use' is replaced with 'no'. This makes no difference for -m.
612 A little builtin syntactic sugar means you can also say
614 -mMODULE=foo,bar or -MMODULE=foo,bar as a shortcut for '-MMODULE
615 qw(foo bar)'. This avoids the need to use quotes when importing
616 symbols. The actual code generated by -MMODULE=foo,bar is "use
617 module split(/,/,q{foo,bar})". Note that the "=" form removes the
618 distinction between -m and -M; that is, -mMODULE=foo,bar is the
619 same as -MMODULE=foo,bar.
620 A consequence of the "split" formulation is that -MMODULE=number
622 never does a version check, unless "MODULE::import()" itself is
623 set up to do a version check, which could happen for example if
624 MODULE inherits from Exporter.
625 -n causes Perl to assume the following loop around your program,
627 which makes it iterate over filename arguments somewhat like sed
628 -n or awk:
629 LINE:
631 while (<>) {
632 ... # your program goes here
633 }
634 Note that the lines are not printed by default. See "-p" to have
636 lines printed. If a file named by an argument cannot be opened
637 for some reason, Perl warns you about it and moves on to the next
638 file.
639 Also note that "<>" passes command line arguments to "open" in
641 perlfunc, which doesn't necessarily interpret them as file names.
642 See perlop for possible security implications.
643 Here is an efficient way to delete all files that haven't been
645 modified for at least a week:
646 find . -mtime +7 -print | perl -nle unlink
648 This is faster than using the -exec switch of find because you
650 don't have to start a process on every filename found (but it's
651 not faster than using the -delete switch available in newer
652 versions of find. It does suffer from the bug of mishandling
653 newlines in pathnames, which you can fix if you follow the example
654 under -0.
655 "BEGIN" and "END" blocks may be used to capture control before or
657 after the implicit program loop, just as in awk.
658 -p causes Perl to assume the following loop around your program,
660 which makes it iterate over filename arguments somewhat like sed:
661 LINE:
663 while (<>) {
664 ... # your program goes here
665 } continue {
666 print or die "-p destination: $!\n";
667 }
668 If a file named by an argument cannot be opened for some reason,
670 Perl warns you about it, and moves on to the next file. Note that
671 the lines are printed automatically. An error occurring during
672 printing is treated as fatal. To suppress printing use the "-n"
673 switch. A -p overrides a -n switch.
674 "BEGIN" and "END" blocks may be used to capture control before or
676 after the implicit loop, just as in awk.
677 -s enables rudimentary switch parsing for switches on the command
679 line after the program name but before any filename arguments (or
680 before an argument of --). Any switch found there is removed from
681 @ARGV and sets the corresponding variable in the Perl program.
682 The following program prints "1" if the program is invoked with a
683 -xyz switch, and "abc" if it is invoked with -xyz=abc.
684 #!/usr/bin/perl -s
686 if ($xyz) { print "$xyz\n" }
687 Do note that a switch like --help creates the variable "${-help}",
689 which is not compliant with "use strict "refs"". Also, when using
690 this option on a script with warnings enabled you may get a lot of
691 spurious "used only once" warnings.
692 -S makes Perl use the "PATH" environment variable to search for the
694 program unless the name of the program contains path separators.
695 On some platforms, this also makes Perl append suffixes to the
697 filename while searching for it. For example, on Win32 platforms,
698 the ".bat" and ".cmd" suffixes are appended if a lookup for the
699 original name fails, and if the name does not already end in one
700 of those suffixes. If your Perl was compiled with "DEBUGGING"
701 turned on, using the -Dp switch to Perl shows how the search
702 progresses.
703 Typically this is used to emulate "#!" startup on platforms that
705 don't support "#!". It's also convenient when debugging a script
706 that uses "#!", and is thus normally found by the shell's $PATH
707 search mechanism.
708 This example works on many platforms that have a shell compatible
710 with Bourne shell:
711 #!/usr/bin/perl
713 eval 'exec /usr/bin/perl -wS $0 ${1+"$@"}'
714 if $running_under_some_shell;
715 The system ignores the first line and feeds the program to
717 /bin/sh, which proceeds to try to execute the Perl program as a
718 shell script. The shell executes the second line as a normal
719 shell command, and thus starts up the Perl interpreter. On some
720 systems $0 doesn't always contain the full pathname, so the "-S"
721 tells Perl to search for the program if necessary. After Perl
722 locates the program, it parses the lines and ignores them because
723 the variable $running_under_some_shell is never true. If the
724 program will be interpreted by csh, you will need to replace
725 "${1+"$@"}" with $*, even though that doesn't understand embedded
726 spaces (and such) in the argument list. To start up sh rather
727 than csh, some systems may have to replace the "#!" line with a
728 line containing just a colon, which will be politely ignored by
729 Perl. Other systems can't control that, and need a totally
730 devious construct that will work under any of csh, sh, or Perl,
731 such as the following:
732 eval '(exit $?0)' && eval 'exec perl -wS $0 ${1+"$@"}'
734 & eval 'exec /usr/bin/perl -wS $0 $argv:q'
735 if $running_under_some_shell;
736 If the filename supplied contains directory separators (and so is
738 an absolute or relative pathname), and if that file is not found,
739 platforms that append file extensions will do so and try to look
740 for the file with those extensions added, one by one.
741 On DOS-like platforms, if the program does not contain directory
743 separators, it will first be searched for in the current directory
744 before being searched for on the PATH. On Unix platforms, the
745 program will be searched for strictly on the PATH.
746 -t Like "-T", but taint checks will issue warnings rather than fatal
748 errors. These warnings can now be controlled normally with "no
749 warnings qw(taint)".
750 Note: This is not a substitute for "-T"! This is meant to be used
752 only as a temporary development aid while securing legacy code:
753 for real production code and for new secure code written from
754 scratch, always use the real "-T".
755 -T turns on "taint" so you can test them. Ordinarily these checks
757 are done only when running setuid or setgid. It's a good idea to
758 turn them on explicitly for programs that run on behalf of someone
759 else whom you might not necessarily trust, such as CGI programs or
760 any internet servers you might write in Perl. See perlsec for
761 details. For security reasons, this option must be seen by Perl
762 quite early; usually this means it must appear early on the
763 command line or in the "#!" line for systems which support that
764 construct.
765 -u This switch causes Perl to dump core after compiling your program.
767 You can then in theory take this core dump and turn it into an
768 executable file by using the undump program (not supplied). This
769 speeds startup at the expense of some disk space (which you can
770 minimize by stripping the executable). (Still, a "hello world"
771 executable comes out to about 200K on my machine.) If you want to
772 execute a portion of your program before dumping, use the
773 "CORE::dump()" function instead. Note: availability of undump is
774 platform specific and may not be available for a specific port of
775 Perl.
776 -U allows Perl to do unsafe operations. Currently the only "unsafe"
778 operations are attempting to unlink directories while running as
779 superuser and running setuid programs with fatal taint checks
780 turned into warnings. Note that warnings must be enabled along
781 with this option to actually generate the taint-check warnings.
782 -v prints the version and patchlevel of your perl executable.
784 -V prints summary of the major perl configuration values and the
786 current values of @INC.
787 -V:configvar
789 Prints to STDOUT the value of the named configuration variable(s),
790 with multiples when your "configvar" argument looks like a regex
791 (has non-letters). For example:
792 $ perl -V:libc
794 libc='/lib/libc-2.2.4.so';
795 $ perl -V:lib.
796 libs='-lnsl -lgdbm -ldb -ldl -lm -lcrypt -lutil -lc';
797 libc='/lib/libc-2.2.4.so';
798 $ perl -V:lib.*
799 libpth='/usr/local/lib /lib /usr/lib';
800 libs='-lnsl -lgdbm -ldb -ldl -lm -lcrypt -lutil -lc';
801 lib_ext='.a';
802 libc='/lib/libc-2.2.4.so';
803 libperl='libperl.a';
804 ....
805 Additionally, extra colons can be used to control formatting. A
807 trailing colon suppresses the linefeed and terminator ";",
808 allowing you to embed queries into shell commands. (mnemonic:
809 PATH separator ":".)
810 $ echo "compression-vars: " `perl -V:z.*: ` " are here !"
812 compression-vars: zcat='' zip='zip' are here !
813 A leading colon removes the "name=" part of the response, this
815 allows you to map to the name you need. (mnemonic: empty label)
816 $ echo "goodvfork="`./perl -Ilib -V::usevfork`
818 goodvfork=false;
819 Leading and trailing colons can be used together if you need
821 positional parameter values without the names. Note that in the
822 case below, the "PERL_API" params are returned in alphabetical
823 order.
824 $ echo building_on `perl -V::osname: -V::PERL_API_.*:` now
826 building_on 'linux' '5' '1' '9' now
827 -w prints warnings about dubious constructs, such as variable names
829 mentioned only once and scalar variables used before being set;
830 redefined subroutines; references to undefined filehandles;
831 filehandles opened read-only that you are attempting to write on;
832 values used as a number that don't look like numbers; using an
833 array as though it were a scalar; if your subroutines recurse more
834 than 100 deep; and innumerable other things.
835 This switch really just enables the global $^W variable; normally,
837 the lexically scoped "use warnings" pragma is preferred. You can
838 disable or promote into fatal errors specific warnings using
839 "__WARN__" hooks, as described in perlvar and "warn" in perlfunc.
840 See also perldiag and perltrap. A fine-grained warning facility
841 is also available if you want to manipulate entire classes of
842 warnings; see warnings.
843 -W Enables all warnings regardless of "no warnings" or $^W. See
845 warnings.
846 -X Disables all warnings regardless of "use warnings" or $^W. See
848 warnings.
849 Forbidden in ""PERL5OPT"".
851 -x
853 -xdirectory
854 tells Perl that the program is embedded in a larger chunk of
855 unrelated text, such as in a mail message. Leading garbage will
856 be discarded until the first line that starts with "#!" and
857 contains the string "perl". Any meaningful switches on that line
858 will be applied.
859 All references to line numbers by the program (warnings, errors,
861 ...) will treat the "#!" line as the first line. Thus a warning
862 on the 2nd line of the program, which is on the 100th line in the
863 file will be reported as line 2, not as line 100. This can be
864 overridden by using the "#line" directive. (See "Plain Old
865 Comments (Not!)" in perlsyn)
866 If a directory name is specified, Perl will switch to that
868 directory before running the program. The -x switch controls only
869 the disposal of leading garbage. The program must be terminated
870 with "__END__" if there is trailing garbage to be ignored; the
871 program can process any or all of the trailing garbage via the
872 "DATA" filehandle if desired.
873 The directory, if specified, must appear immediately following the
875 -x with no intervening whitespace.
876
878 HOME Used if "chdir" has no argument.
879 LOGDIR Used if "chdir" has no argument and "HOME" is not set.
881 PATH Used in executing subprocesses, and in finding the program
883 if "-S" is used.
884 PERL5LIB A list of directories in which to look for Perl library
886 files before looking in the standard library. Any
887 architecture-specific and version-specific directories,
888 such as version/archname/, version/, or archname/ under the
889 specified locations are automatically included if they
890 exist, with this lookup done at interpreter startup time.
891 In addition, any directories matching the entries in
892 $Config{inc_version_list} are added. (These typically
893 would be for older compatible perl versions installed in
894 the same directory tree.)
895 If PERL5LIB is not defined, "PERLLIB" is used. Directories
897 are separated (like in PATH) by a colon on Unixish
898 platforms and by a semicolon on Windows (the proper path
899 separator being given by the command "perl -V:path_sep").
900 When running taint checks, either because the program was
902 running setuid or setgid, or the "-T" or "-t" switch was
903 specified, neither PERL5LIB nor "PERLLIB" is consulted. The
904 program should instead say:
905 use lib "/my/directory";
907 PERL5OPT Command-line options (switches). Switches in this variable
909 are treated as if they were on every Perl command line.
910 Only the -[CDIMTUWdmtw] switches are allowed. When running
911 taint checks (either because the program was running setuid
912 or setgid, or because the "-T" or "-t" switch was used),
913 this variable is ignored. If PERL5OPT begins with -T,
914 tainting will be enabled and subsequent options ignored.
915 If PERL5OPT begins with -t, tainting will be enabled, a
916 writable dot removed from @INC, and subsequent options
917 honored.
918 PERLIO A space (or colon) separated list of PerlIO layers. If perl
920 is built to use PerlIO system for IO (the default) these
921 layers affect Perl's IO.
922 It is conventional to start layer names with a colon (for
924 example, ":perlio") to emphasize their similarity to
925 variable "attributes". But the code that parses layer
926 specification strings, which is also used to decode the
927 PERLIO environment variable, treats the colon as a
928 separator.
929 An unset or empty PERLIO is equivalent to the default set
931 of layers for your platform; for example, ":unix:perlio" on
932 Unix-like systems and ":unix:crlf" on Windows and other
933 DOS-like systems.
934 The list becomes the default for all Perl's IO.
936 Consequently only built-in layers can appear in this list,
937 as external layers (such as ":encoding()") need IO in order
938 to load them! See "open pragma" for how to add external
939 encodings as defaults.
940 Layers it makes sense to include in the PERLIO environment
942 variable are briefly summarized below. For more details see
943 PerlIO.
944 :crlf A layer which does CRLF to "\n" translation
946 distinguishing "text" and "binary" files in the
947 manner of MS-DOS and similar operating systems, and
948 also provides buffering similar to ":perlio" on
949 these architectures.
950 :perlio This is a re-implementation of stdio-like buffering
952 written as a PerlIO layer. As such it will call
953 whatever layer is below it for its operations,
954 typically ":unix".
955 :stdio This layer provides a PerlIO interface by wrapping
957 system's ANSI C "stdio" library calls. The layer
958 provides both buffering and IO. Note that the
959 ":stdio" layer does not do CRLF translation even if
960 that is the platform's normal behaviour. You will
961 need a ":crlf" layer above it to do that.
962 :unix Low-level layer that calls "read", "write",
964 "lseek", etc.
965 :win32 On Win32 platforms this experimental layer uses
967 native "handle" IO rather than a Unix-like numeric
968 file descriptor layer. Known to be buggy in this
969 release (5.30).
970 The default set of layers should give acceptable results on
972 all platforms.
973 For Unix platforms that will be the equivalent of
975 ":unix:perlio" or ":stdio". Configure is set up to prefer
976 the ":stdio" implementation if the system's library
977 provides for fast access to the buffer (not common on
978 modern architectures); otherwise, it uses the
979 ":unix:perlio" implementation.
980 On Win32 the default in this release (5.30) is
982 ":unix:crlf". Win32's ":stdio" has a number of
983 bugs/mis-features for Perl IO which are somewhat depending
984 on the version and vendor of the C compiler. Using our own
985 ":crlf" layer as the buffer avoids those issues and makes
986 things more uniform.
987 This release (5.30) uses ":unix" as the bottom layer on
989 Win32, and so still uses the C compiler's numeric file
990 descriptor routines. There is an experimental native
991 ":win32" layer, which is expected to be enhanced and may
992 eventually become the default under Win32.
993 The PERLIO environment variable is completely ignored when
995 Perl is run in taint mode.
996 PERLIO_DEBUG
998 If set to the name of a file or device when Perl is run
999 with the -Di command-line switch, the logging of certain
1000 operations of the PerlIO subsystem will be redirected to
1001 the specified file rather than going to stderr, which is
1002 the default. The file is opened in append mode. Typical
1003 uses are in Unix:
1004 % env PERLIO_DEBUG=/tmp/perlio.log perl -Di script ...
1006 and under Win32, the approximately equivalent:
1008 > set PERLIO_DEBUG=CON
1010 perl -Di script ...
1011 This functionality is disabled for setuid scripts, for
1013 scripts run with "-T", and for scripts run on a Perl built
1014 without "-DDEBUGGING" support.
1015 PERLLIB A list of directories in which to look for Perl library
1017 files before looking in the standard library. If
1018 "PERL5LIB" is defined, PERLLIB is not used.
1019 The PERLLIB environment variable is completely ignored when
1021 Perl is run in taint mode.
1022 PERL5DB The command used to load the debugger code. The default
1024 is:
1025 BEGIN { require "perl5db.pl" }
1027 The PERL5DB environment variable is only used when Perl is
1029 started with a bare "-d" switch.
1030 PERL5DB_THREADED
1032 If set to a true value, indicates to the debugger that the
1033 code being debugged uses threads.
1034 PERL5SHELL (specific to the Win32 port)
1036 On Win32 ports only, may be set to an alternative shell
1037 that Perl must use internally for executing "backtick"
1038 commands or system(). Default is "cmd.exe /x/d/c" on
1039 WindowsNT and "command.com /c" on Windows95. The value is
1040 considered space-separated. Precede any character that
1041 needs to be protected, like a space or backslash, with
1042 another backslash.
1043 Note that Perl doesn't use COMSPEC for this purpose because
1045 COMSPEC has a high degree of variability among users,
1046 leading to portability concerns. Besides, Perl can use a
1047 shell that may not be fit for interactive use, and setting
1048 COMSPEC to such a shell may interfere with the proper
1049 functioning of other programs (which usually look in
1050 COMSPEC to find a shell fit for interactive use).
1051 Before Perl 5.10.0 and 5.8.8, PERL5SHELL was not taint
1053 checked when running external commands. It is recommended
1054 that you explicitly set (or delete) $ENV{PERL5SHELL} when
1055 running in taint mode under Windows.
1056 PERL_ALLOW_NON_IFS_LSP (specific to the Win32 port)
1058 Set to 1 to allow the use of non-IFS compatible LSPs
1059 (Layered Service Providers). Perl normally searches for an
1060 IFS-compatible LSP because this is required for its
1061 emulation of Windows sockets as real filehandles. However,
1062 this may cause problems if you have a firewall such as
1063 McAfee Guardian, which requires that all applications use
1064 its LSP but which is not IFS-compatible, because clearly
1065 Perl will normally avoid using such an LSP.
1066 Setting this environment variable to 1 means that Perl will
1068 simply use the first suitable LSP enumerated in the
1069 catalog, which keeps McAfee Guardian happy--and in that
1070 particular case Perl still works too because McAfee
1071 Guardian's LSP actually plays other games which allow
1072 applications requiring IFS compatibility to work.
1073 PERL_DEBUG_MSTATS
1075 Relevant only if Perl is compiled with the "malloc"
1076 included with the Perl distribution; that is, if "perl
1077 -V:d_mymalloc" is "define".
1078 If set, this dumps out memory statistics after execution.
1080 If set to an integer greater than one, also dumps out
1081 memory statistics after compilation.
1082 PERL_DESTRUCT_LEVEL
1084 Controls the behaviour of global destruction of objects and
1085 other references. See "PERL_DESTRUCT_LEVEL" in
1086 perlhacktips for more information.
1087 PERL_DL_NONLAZY
1089 Set to "1" to have Perl resolve all undefined symbols when
1090 it loads a dynamic library. The default behaviour is to
1091 resolve symbols when they are used. Setting this variable
1092 is useful during testing of extensions, as it ensures that
1093 you get an error on misspelled function names even if the
1094 test suite doesn't call them.
1095 PERL_ENCODING
1097 If using the "use encoding" pragma without an explicit
1098 encoding name, the PERL_ENCODING environment variable is
1099 consulted for an encoding name.
1100 PERL_HASH_SEED
1102 (Since Perl 5.8.1, new semantics in Perl 5.18.0) Used to
1103 override the randomization of Perl's internal hash
1104 function. The value is expressed in hexadecimal, and may
1105 include a leading 0x. Truncated patterns are treated as
1106 though they are suffixed with sufficient 0's as required.
1107 If the option is provided, and "PERL_PERTURB_KEYS" is NOT
1109 set, then a value of '0' implies "PERL_PERTURB_KEYS=0" and
1110 any other value implies "PERL_PERTURB_KEYS=2".
1111 PLEASE NOTE: The hash seed is sensitive information. Hashes
1113 are randomized to protect against local and remote attacks
1114 against Perl code. By manually setting a seed, this
1115 protection may be partially or completely lost.
1116 See "Algorithmic Complexity Attacks" in perlsec,
1118 "PERL_PERTURB_KEYS", and "PERL_HASH_SEED_DEBUG" for more
1119 information.
1120 PERL_PERTURB_KEYS
1122 (Since Perl 5.18.0) Set to "0" or "NO" then traversing
1123 keys will be repeatable from run to run for the same
1124 "PERL_HASH_SEED". Insertion into a hash will not change
1125 the order, except to provide for more space in the hash.
1126 When combined with setting PERL_HASH_SEED this mode is as
1127 close to pre 5.18 behavior as you can get.
1128 When set to "1" or "RANDOM" then traversing keys will be
1130 randomized. Every time a hash is inserted into the key
1131 order will change in a random fashion. The order may not be
1132 repeatable in a following program run even if the
1133 PERL_HASH_SEED has been specified. This is the default mode
1134 for perl.
1135 When set to "2" or "DETERMINISTIC" then inserting keys into
1137 a hash will cause the key order to change, but in a way
1138 that is repeatable from program run to program run.
1139 NOTE: Use of this option is considered insecure, and is
1141 intended only for debugging non-deterministic behavior in
1142 Perl's hash function. Do not use it in production.
1143 See "Algorithmic Complexity Attacks" in perlsec and
1145 "PERL_HASH_SEED" and "PERL_HASH_SEED_DEBUG" for more
1146 information. You can get and set the key traversal mask for
1147 a specific hash by using the "hash_traversal_mask()"
1148 function from Hash::Util.
1149 PERL_HASH_SEED_DEBUG
1151 (Since Perl 5.8.1.) Set to "1" to display (to STDERR)
1152 information about the hash function, seed, and what type of
1153 key traversal randomization is in effect at the beginning
1154 of execution. This, combined with "PERL_HASH_SEED" and
1155 "PERL_PERTURB_KEYS" is intended to aid in debugging
1156 nondeterministic behaviour caused by hash randomization.
1157 Note that any information about the hash function,
1159 especially the hash seed is sensitive information: by
1160 knowing it, one can craft a denial-of-service attack
1161 against Perl code, even remotely; see "Algorithmic
1162 Complexity Attacks" in perlsec for more information. Do not
1163 disclose the hash seed to people who don't need to know it.
1164 See also "hash_seed()" and "hash_traversal_mask()".
1165 An example output might be:
1167 HASH_FUNCTION = ONE_AT_A_TIME_HARD HASH_SEED = 0x652e9b9349a7a032 PERTURB_KEYS = 1 (RANDOM)
1169 PERL_MEM_LOG
1171 If your Perl was configured with -Accflags=-DPERL_MEM_LOG,
1172 setting the environment variable "PERL_MEM_LOG" enables
1173 logging debug messages. The value has the form
1174 "<number>[m][s][t]", where "number" is the file descriptor
1175 number you want to write to (2 is default), and the
1176 combination of letters specifies that you want information
1177 about (m)emory and/or (s)v, optionally with (t)imestamps.
1178 For example, "PERL_MEM_LOG=1mst" logs all information to
1179 stdout. You can write to other opened file descriptors in a
1180 variety of ways:
1181 $ 3>foo3 PERL_MEM_LOG=3m perl ...
1183 PERL_ROOT (specific to the VMS port)
1185 A translation-concealed rooted logical name that contains
1186 Perl and the logical device for the @INC path on VMS only.
1187 Other logical names that affect Perl on VMS include
1188 PERLSHR, PERL_ENV_TABLES, and SYS$TIMEZONE_DIFFERENTIAL,
1189 but are optional and discussed further in perlvms and in
1190 README.vms in the Perl source distribution.
1191 PERL_SIGNALS
1193 Available in Perls 5.8.1 and later. If set to "unsafe",
1194 the pre-Perl-5.8.0 signal behaviour (which is immediate but
1195 unsafe) is restored. If set to "safe", then safe (but
1196 deferred) signals are used. See "Deferred Signals (Safe
1197 Signals)" in perlipc.
1198 PERL_UNICODE
1200 Equivalent to the -C command-line switch. Note that this
1201 is not a boolean variable. Setting this to "1" is not the
1202 right way to "enable Unicode" (whatever that would mean).
1203 You can use "0" to "disable Unicode", though (or
1204 alternatively unset PERL_UNICODE in your shell before
1205 starting Perl). See the description of the -C switch for
1206 more information.
1207 PERL_USE_UNSAFE_INC
1209 If perl has been configured to not have the current
1210 directory in @INC by default, this variable can be set to
1211 "1" to reinstate it. It's primarily intended for use while
1212 building and testing modules that have not been updated to
1213 deal with "." not being in @INC and should not be set in
1214 the environment for day-to-day use.
1215 SYS$LOGIN (specific to the VMS port)
1217 Used if chdir has no argument and "HOME" and "LOGDIR" are
1218 not set.
1219 PERL_INTERNAL_RAND_SEED
1221 Set to a non-negative integer to seed the random number
1222 generator used internally by perl for a variety of
1223 purposes.
1224 Ignored if perl is run setuid or setgid. Used only for
1226 some limited startup randomization (hash keys) if "-T" or
1227 "-t" perl is started with tainting enabled.
1228 Perl may be built to ignore this variable.
1230 Perl also has environment variables that control how Perl handles data
1232 specific to particular natural languages; see perllocale.
1233 Perl and its various modules and components, including its test
1235 frameworks, may sometimes make use of certain other environment
1236 variables. Some of these are specific to a particular platform.
1237 Please consult the appropriate module documentation and any
1238 documentation for your platform (like perlsolaris, perllinux,
1239 perlmacosx, perlwin32, etc) for variables peculiar to those specific
1240 situations.
1241 Perl makes all environment variables available to the program being
1243 executed, and passes these along to any child processes it starts.
1244 However, programs running setuid would do well to execute the following
1245 lines before doing anything else, just to keep people honest:
1246 $ENV{PATH} = "/bin:/usr/bin"; # or whatever you need
1248 $ENV{SHELL} = "/bin/sh" if exists $ENV{SHELL};
1249 delete @ENV{qw(IFS CDPATH ENV BASH_ENV)};
1250
1252 Some options, in particular "-I", "-M", "PERL5LIB" and "PERL5OPT" can
1253 interact, and the order in which they are applied is important.
1254 Note that this section does not document what actually happens inside
1256 the perl interpreter, it documents what effectively happens.
1257 -I The effect of multiple "-I" options is to "unshift" them onto @INC
1259 from right to left. So for example:
1260 perl -I 1 -I 2 -I 3
1262 will first prepend 3 onto the front of @INC, then prepend 2, and
1264 then prepend 1. The result is that @INC begins with:
1265 qw(1 2 3)
1267 -M Multiple "-M" options are processed from left to right. So this:
1269 perl -Mlib=1 -Mlib=2 -Mlib=3
1271 will first use the lib pragma to prepend 1 to @INC, then it will
1273 prepend 2, then it will prepend 3, resulting in an @INC that begins
1274 with:
1275 qw(3 2 1)
1277 the PERL5LIB environment variable
1279 This contains a list of directories, separated by colons. The
1280 entire list is prepended to @INC in one go. This:
1281 PERL5LIB=1:2:3 perl
1283 will result in an @INC that begins with:
1285 qw(1 2 3)
1287 combinations of -I, -M and PERL5LIB
1289 "PERL5LIB" is applied first, then all the "-I" arguments, then all
1290 the "-M" arguments. This:
1291 PERL5LIB=e1:e2 perl -I i1 -Mlib=m1 -I i2 -Mlib=m2
1293 will result in an @INC that begins with:
1295 qw(m2 m1 i1 i2 e1 e2)
1297 the PERL5OPT environment variable
1299 This contains a space separated list of switches. We only consider
1300 the effects of "-M" and "-I" in this section.
1301 After normal processing of "-I" switches from the command line, all
1303 the "-I" switches in "PERL5OPT" are extracted. They are processed
1304 from left to right instead of from right to left. Also note that
1305 while whitespace is allowed between a "-I" and its directory on the
1306 command line, it is not allowed in "PERL5OPT".
1307 After normal processing of "-M" switches from the command line, all
1309 the "-M" switches in "PERL5OPT" are extracted. They are processed
1310 from left to right, i.e. the same as those on the command line.
1311 An example may make this clearer:
1313 export PERL5OPT="-Mlib=optm1 -Iopti1 -Mlib=optm2 -Iopti2"
1315 export PERL5LIB=e1:e2
1316 perl -I i1 -Mlib=m1 -I i2 -Mlib=m2
1317 will result in an @INC that begins with:
1319 qw(
1321 optm2
1322 optm1
1323 m2
1325 m1
1326 opti2
1328 opti1
1329 i1
1331 i2
1332 e1
1334 e2
1335 )
1336 Other complications
1338 There are some complications that are ignored in the examples
1339 above:
1340 arch and version subdirs
1342 All of "-I", "PERL5LIB" and "use lib" will also prepend arch
1343 and version subdirs if they are present
1344 sitecustomize.pl
1346perl v5.32.1 2021-03-31 PERLRUN(1)