3 perlport - Writing portable Perl
7 Perl runs on numerous operating systems. While most of them share
8 much in common, they also have their own unique features.
10 This document is meant to help you to find out what constitutes portable
11 Perl code. That way once you make a decision to write portably,
12 you know where the lines are drawn, and you can stay within them.
14 There is a tradeoff between taking full advantage of one particular
15 type of computer and taking advantage of a full range of them.
16 Naturally, as you broaden your range and become more diverse, the
17 common factors drop, and you are left with an increasingly smaller
18 area of common ground in which you can operate to accomplish a
19 particular task. Thus, when you begin attacking a problem, it is
20 important to consider under which part of the tradeoff curve you
21 want to operate. Specifically, you must decide whether it is
22 important that the task that you are coding have the full generality
23 of being portable, or whether to just get the job done right now.
24 This is the hardest choice to be made. The rest is easy, because
25 Perl provides many choices, whichever way you want to approach your
28 Looking at it another way, writing portable code is usually about
29 willfully limiting your available choices. Naturally, it takes
30 discipline and sacrifice to do that. The product of portability
31 and convenience may be a constant. You have been warned.
33 Be aware of two important points:
37 =item Not all Perl programs have to be portable
39 There is no reason you should not use Perl as a language to glue Unix
40 tools together, or to prototype a Macintosh application, or to manage the
41 Windows registry. If it makes no sense to aim for portability for one
42 reason or another in a given program, then don't bother.
44 =item Nearly all of Perl already I<is> portable
46 Don't be fooled into thinking that it is hard to create portable Perl
47 code. It isn't. Perl tries its level-best to bridge the gaps between
48 what's available on different platforms, and all the means available to
49 use those features. Thus almost all Perl code runs on any machine
50 without modification. But there are some significant issues in
51 writing portable code, and this document is entirely about those issues.
55 Here's the general rule: When you approach a task commonly done
56 using a whole range of platforms, think about writing portable
57 code. That way, you don't sacrifice much by way of the implementation
58 choices you can avail yourself of, and at the same time you can give
59 your users lots of platform choices. On the other hand, when you have to
60 take advantage of some unique feature of a particular platform, as is
61 often the case with systems programming (whether for Unix, Windows,
62 S<Mac OS>, VMS, etc.), consider writing platform-specific code.
64 When the code will run on only two or three operating systems, you
65 may need to consider only the differences of those particular systems.
66 The important thing is to decide where the code will run and to be
67 deliberate in your decision.
69 The material below is separated into three main sections: main issues of
70 portability (L<"ISSUES">), platform-specific issues (L<"PLATFORMS">), and
71 built-in perl functions that behave differently on various ports
72 (L<"FUNCTION IMPLEMENTATIONS">).
74 This information should not be considered complete; it includes possibly
75 transient information about idiosyncrasies of some of the ports, almost
76 all of which are in a state of constant evolution. Thus, this material
77 should be considered a perpetual work in progress
78 (C<< <IMG SRC="yellow_sign.gif" ALT="Under Construction"> >>).
84 In most operating systems, lines in files are terminated by newlines.
85 Just what is used as a newline may vary from OS to OS. Unix
86 traditionally uses C<\012>, one type of DOSish I/O uses C<\015\012>,
87 and S<Mac OS> uses C<\015>.
89 Perl uses C<\n> to represent the "logical" newline, where what is
90 logical may depend on the platform in use. In MacPerl, C<\n> always
91 means C<\015>. In DOSish perls, C<\n> usually means C<\012>, but
92 when accessing a file in "text" mode, STDIO translates it to (or
93 from) C<\015\012>, depending on whether you're reading or writing.
94 Unix does the same thing on ttys in canonical mode. C<\015\012>
95 is commonly referred to as CRLF.
97 To trim trailing newlines from text lines use chomp(). With default
98 settings that function looks for a trailing C<\n> character and thus
99 trims in a portable way.
101 When dealing with binary files (or text files in binary mode) be sure
102 to explicitly set $/ to the appropriate value for your file format
103 before using chomp().
105 Because of the "text" mode translation, DOSish perls have limitations
106 in using C<seek> and C<tell> on a file accessed in "text" mode.
107 Stick to C<seek>-ing to locations you got from C<tell> (and no
108 others), and you are usually free to use C<seek> and C<tell> even
109 in "text" mode. Using C<seek> or C<tell> or other file operations
110 may be non-portable. If you use C<binmode> on a file, however, you
111 can usually C<seek> and C<tell> with arbitrary values in safety.
113 A common misconception in socket programming is that C<\n> eq C<\012>
114 everywhere. When using protocols such as common Internet protocols,
115 C<\012> and C<\015> are called for specifically, and the values of
116 the logical C<\n> and C<\r> (carriage return) are not reliable.
118 print SOCKET "Hi there, client!\r\n"; # WRONG
119 print SOCKET "Hi there, client!\015\012"; # RIGHT
121 However, using C<\015\012> (or C<\cM\cJ>, or C<\x0D\x0A>) can be tedious
122 and unsightly, as well as confusing to those maintaining the code. As
123 such, the Socket module supplies the Right Thing for those who want it.
125 use Socket qw(:DEFAULT :crlf);
126 print SOCKET "Hi there, client!$CRLF" # RIGHT
128 When reading from a socket, remember that the default input record
129 separator C<$/> is C<\n>, but robust socket code will recognize as
130 either C<\012> or C<\015\012> as end of line:
136 Because both CRLF and LF end in LF, the input record separator can
137 be set to LF and any CR stripped later. Better to write:
139 use Socket qw(:DEFAULT :crlf);
140 local($/) = LF; # not needed if $/ is already \012
143 s/$CR?$LF/\n/; # not sure if socket uses LF or CRLF, OK
144 # s/\015?\012/\n/; # same thing
147 This example is preferred over the previous one--even for Unix
148 platforms--because now any C<\015>'s (C<\cM>'s) are stripped out
149 (and there was much rejoicing).
151 Similarly, functions that return text data--such as a function that
152 fetches a web page--should sometimes translate newlines before
153 returning the data, if they've not yet been translated to the local
154 newline representation. A single line of code will often suffice:
156 $data =~ s/\015?\012/\n/g;
159 Some of this may be confusing. Here's a handy reference to the ASCII CR
160 and LF characters. You can print it out and stick it in your wallet.
162 LF eq \012 eq \x0A eq \cJ eq chr(10) eq ASCII 10
163 CR eq \015 eq \x0D eq \cM eq chr(13) eq ASCII 13
166 ---------------------------
169 \n * | LF | CRLF | CR |
170 \r * | CR | CR | LF |
171 ---------------------------
174 The Unix column assumes that you are not accessing a serial line
175 (like a tty) in canonical mode. If you are, then CR on input becomes
176 "\n", and "\n" on output becomes CRLF.
178 These are just the most common definitions of C<\n> and C<\r> in Perl.
179 There may well be others. For example, on an EBCDIC implementation
180 such as z/OS (OS/390) or OS/400 (using the ILE, the PASE is ASCII-based)
181 the above material is similar to "Unix" but the code numbers change:
183 LF eq \025 eq \x15 eq \cU eq chr(21) eq CP-1047 21
184 LF eq \045 eq \x25 eq chr(37) eq CP-0037 37
185 CR eq \015 eq \x0D eq \cM eq chr(13) eq CP-1047 13
186 CR eq \015 eq \x0D eq \cM eq chr(13) eq CP-0037 13
189 ----------------------
194 ----------------------
197 =head2 Numbers endianness and Width
199 Different CPUs store integers and floating point numbers in different
200 orders (called I<endianness>) and widths (32-bit and 64-bit being the
201 most common today). This affects your programs when they attempt to transfer
202 numbers in binary format from one CPU architecture to another,
203 usually either "live" via network connection, or by storing the
204 numbers to secondary storage such as a disk file or tape.
206 Conflicting storage orders make utter mess out of the numbers. If a
207 little-endian host (Intel, VAX) stores 0x12345678 (305419896 in
208 decimal), a big-endian host (Motorola, Sparc, PA) reads it as
209 0x78563412 (2018915346 in decimal). Alpha and MIPS can be either:
210 Digital/Compaq used/uses them in little-endian mode; SGI/Cray uses
211 them in big-endian mode. To avoid this problem in network (socket)
212 connections use the C<pack> and C<unpack> formats C<n> and C<N>, the
213 "network" orders. These are guaranteed to be portable.
215 As of perl 5.9.2, you can also use the C<E<gt>> and C<E<lt>> modifiers
216 to force big- or little-endian byte-order. This is useful if you want
217 to store signed integers or 64-bit integers, for example.
219 You can explore the endianness of your platform by unpacking a
220 data structure packed in native format such as:
222 print unpack("h*", pack("s2", 1, 2)), "\n";
223 # '10002000' on e.g. Intel x86 or Alpha 21064 in little-endian mode
224 # '00100020' on e.g. Motorola 68040
226 If you need to distinguish between endian architectures you could use
227 either of the variables set like so:
229 $is_big_endian = unpack("h*", pack("s", 1)) =~ /01/;
230 $is_little_endian = unpack("h*", pack("s", 1)) =~ /^1/;
232 Differing widths can cause truncation even between platforms of equal
233 endianness. The platform of shorter width loses the upper parts of the
234 number. There is no good solution for this problem except to avoid
235 transferring or storing raw binary numbers.
237 One can circumnavigate both these problems in two ways. Either
238 transfer and store numbers always in text format, instead of raw
239 binary, or else consider using modules like Data::Dumper (included in
240 the standard distribution as of Perl 5.005) and Storable (included as
241 of perl 5.8). Keeping all data as text significantly simplifies matters.
243 The v-strings are portable only up to v2147483647 (0x7FFFFFFF), that's
244 how far EBCDIC, or more precisely UTF-EBCDIC will go.
246 =head2 Files and Filesystems
248 Most platforms these days structure files in a hierarchical fashion.
249 So, it is reasonably safe to assume that all platforms support the
250 notion of a "path" to uniquely identify a file on the system. How
251 that path is really written, though, differs considerably.
253 Although similar, file path specifications differ between Unix,
254 Windows, S<Mac OS>, OS/2, VMS, VOS, S<RISC OS>, and probably others.
255 Unix, for example, is one of the few OSes that has the elegant idea
256 of a single root directory.
258 DOS, OS/2, VMS, VOS, and Windows can work similarly to Unix with C</>
259 as path separator, or in their own idiosyncratic ways (such as having
260 several root directories and various "unrooted" device files such NIL:
263 S<Mac OS> uses C<:> as a path separator instead of C</>.
265 The filesystem may support neither hard links (C<link>) nor
266 symbolic links (C<symlink>, C<readlink>, C<lstat>).
268 The filesystem may support neither access timestamp nor change
269 timestamp (meaning that about the only portable timestamp is the
270 modification timestamp), or one second granularity of any timestamps
271 (e.g. the FAT filesystem limits the time granularity to two seconds).
273 The "inode change timestamp" (the C<-C> filetest) may really be the
274 "creation timestamp" (which it is not in UNIX).
276 VOS perl can emulate Unix filenames with C</> as path separator. The
277 native pathname characters greater-than, less-than, number-sign, and
278 percent-sign are always accepted.
280 S<RISC OS> perl can emulate Unix filenames with C</> as path
281 separator, or go native and use C<.> for path separator and C<:> to
282 signal filesystems and disk names.
284 Don't assume UNIX filesystem access semantics: that read, write,
285 and execute are all the permissions there are, and even if they exist,
286 that their semantics (for example what do r, w, and x mean on
287 a directory) are the UNIX ones. The various UNIX/POSIX compatibility
288 layers usually try to make interfaces like chmod() work, but sometimes
289 there simply is no good mapping.
291 If all this is intimidating, have no (well, maybe only a little)
292 fear. There are modules that can help. The File::Spec modules
293 provide methods to do the Right Thing on whatever platform happens
294 to be running the program.
296 use File::Spec::Functions;
297 chdir(updir()); # go up one directory
298 $file = catfile(curdir(), 'temp', 'file.txt');
299 # on Unix and Win32, './temp/file.txt'
300 # on Mac OS, ':temp:file.txt'
301 # on VMS, '[.temp]file.txt'
303 File::Spec is available in the standard distribution as of version
304 5.004_05. File::Spec::Functions is only in File::Spec 0.7 and later,
305 and some versions of perl come with version 0.6. If File::Spec
306 is not updated to 0.7 or later, you must use the object-oriented
307 interface from File::Spec (or upgrade File::Spec).
309 In general, production code should not have file paths hardcoded.
310 Making them user-supplied or read from a configuration file is
311 better, keeping in mind that file path syntax varies on different
314 This is especially noticeable in scripts like Makefiles and test suites,
315 which often assume C</> as a path separator for subdirectories.
317 Also of use is File::Basename from the standard distribution, which
318 splits a pathname into pieces (base filename, full path to directory,
321 Even when on a single platform (if you can call Unix a single platform),
322 remember not to count on the existence or the contents of particular
323 system-specific files or directories, like F</etc/passwd>,
324 F</etc/sendmail.conf>, F</etc/resolv.conf>, or even F</tmp/>. For
325 example, F</etc/passwd> may exist but not contain the encrypted
326 passwords, because the system is using some form of enhanced security.
327 Or it may not contain all the accounts, because the system is using NIS.
328 If code does need to rely on such a file, include a description of the
329 file and its format in the code's documentation, then make it easy for
330 the user to override the default location of the file.
332 Don't assume a text file will end with a newline. They should,
335 Do not have two files or directories of the same name with different
336 case, like F<test.pl> and F<Test.pl>, as many platforms have
337 case-insensitive (or at least case-forgiving) filenames. Also, try
338 not to have non-word characters (except for C<.>) in the names, and
339 keep them to the 8.3 convention, for maximum portability, onerous a
340 burden though this may appear.
342 Likewise, when using the AutoSplit module, try to keep your functions to
343 8.3 naming and case-insensitive conventions; or, at the least,
344 make it so the resulting files have a unique (case-insensitively)
347 Whitespace in filenames is tolerated on most systems, but not all,
348 and even on systems where it might be tolerated, some utilities
349 might become confused by such whitespace.
351 Many systems (DOS, VMS ODS-2) cannot have more than one C<.> in their
354 Don't assume C<< > >> won't be the first character of a filename.
355 Always use C<< < >> explicitly to open a file for reading, or even
356 better, use the three-arg version of open, unless you want the user to
357 be able to specify a pipe open.
359 open(FILE, '<', $existing_file) or die $!;
361 If filenames might use strange characters, it is safest to open it
362 with C<sysopen> instead of C<open>. C<open> is magic and can
363 translate characters like C<< > >>, C<< < >>, and C<|>, which may
364 be the wrong thing to do. (Sometimes, though, it's the right thing.)
365 Three-arg open can also help protect against this translation in cases
366 where it is undesirable.
368 Don't use C<:> as a part of a filename since many systems use that for
369 their own semantics (Mac OS Classic for separating pathname components,
370 many networking schemes and utilities for separating the nodename and
371 the pathname, and so on). For the same reasons, avoid C<@>, C<;> and
374 Don't assume that in pathnames you can collapse two leading slashes
375 C<//> into one: some networking and clustering filesystems have special
376 semantics for that. Let the operating system to sort it out.
378 The I<portable filename characters> as defined by ANSI C are
380 a b c d e f g h i j k l m n o p q r t u v w x y z
381 A B C D E F G H I J K L M N O P Q R T U V W X Y Z
385 and the "-" shouldn't be the first character. If you want to be
386 hypercorrect, stay case-insensitive and within the 8.3 naming
387 convention (all the files and directories have to be unique within one
388 directory if their names are lowercased and truncated to eight
389 characters before the C<.>, if any, and to three characters after the
390 C<.>, if any). (And do not use C<.>s in directory names.)
392 =head2 System Interaction
394 Not all platforms provide a command line. These are usually platforms
395 that rely primarily on a Graphical User Interface (GUI) for user
396 interaction. A program requiring a command line interface might
397 not work everywhere. This is probably for the user of the program
398 to deal with, so don't stay up late worrying about it.
400 Some platforms can't delete or rename files held open by the system,
401 this limitation may also apply to changing filesystem metainformation
402 like file permissions or owners. Remember to C<close> files when you
403 are done with them. Don't C<unlink> or C<rename> an open file. Don't
404 C<tie> or C<open> a file already tied or opened; C<untie> or C<close>
407 Don't open the same file more than once at a time for writing, as some
408 operating systems put mandatory locks on such files.
410 Don't assume that write/modify permission on a directory gives the
411 right to add or delete files/directories in that directory. That is
412 filesystem specific: in some filesystems you need write/modify
413 permission also (or even just) in the file/directory itself. In some
414 filesystems (AFS, DFS) the permission to add/delete directory entries
415 is a completely separate permission.
417 Don't assume that a single C<unlink> completely gets rid of the file:
418 some filesystems (most notably the ones in VMS) have versioned
419 filesystems, and unlink() removes only the most recent one (it doesn't
420 remove all the versions because by default the native tools on those
421 platforms remove just the most recent version, too). The portable
422 idiom to remove all the versions of a file is
424 1 while unlink "file";
426 This will terminate if the file is undeleteable for some reason
427 (protected, not there, and so on).
429 Don't count on a specific environment variable existing in C<%ENV>.
430 Don't count on C<%ENV> entries being case-sensitive, or even
431 case-preserving. Don't try to clear %ENV by saying C<%ENV = ();>, or,
432 if you really have to, make it conditional on C<$^O ne 'VMS'> since in
433 VMS the C<%ENV> table is much more than a per-process key-value string
436 On VMS, some entries in the %ENV hash are dynamically created when
437 their key is used on a read if they did not previously exist. The
438 values for C<$ENV{HOME}>, C<$ENV{TERM}>, C<$ENV{HOME}>, and C<$ENV{USER}>,
439 are known to be dynamically generated. The specific names that are
440 dynamically generated may vary with the version of the C library on VMS,
441 and more may exist than is documented.
443 On VMS by default, changes to the %ENV hash are persistent after the process
444 exits. This can cause unintended issues.
446 Don't count on signals or C<%SIG> for anything.
448 Don't count on filename globbing. Use C<opendir>, C<readdir>, and
451 Don't count on per-program environment variables, or per-program current
454 Don't count on specific values of C<$!>, neither numeric nor
455 especially the strings values-- users may switch their locales causing
456 error messages to be translated into their languages. If you can
457 trust a POSIXish environment, you can portably use the symbols defined
458 by the Errno module, like ENOENT. And don't trust on the values of C<$!>
459 at all except immediately after a failed system call.
461 =head2 Command names versus file pathnames
463 Don't assume that the name used to invoke a command or program with
464 C<system> or C<exec> can also be used to test for the existence of the
465 file that holds the executable code for that command or program.
466 First, many systems have "internal" commands that are built-in to the
467 shell or OS and while these commands can be invoked, there is no
468 corresponding file. Second, some operating systems (e.g., Cygwin,
469 DJGPP, OS/2, and VOS) have required suffixes for executable files;
470 these suffixes are generally permitted on the command name but are not
471 required. Thus, a command like "perl" might exist in a file named
472 "perl", "perl.exe", or "perl.pm", depending on the operating system.
473 The variable "_exe" in the Config module holds the executable suffix,
474 if any. Third, the VMS port carefully sets up $^X and
475 $Config{perlpath} so that no further processing is required. This is
476 just as well, because the matching regular expression used below would
477 then have to deal with a possible trailing version number in the VMS
480 To convert $^X to a file pathname, taking account of the requirements
481 of the various operating system possibilities, say:
486 {$thisperl .= $Config{_exe} unless $thisperl =~ m/$Config{_exe}$/i;}
488 To convert $Config{perlpath} to a file pathname, say:
491 $thisperl = $Config{perlpath};
493 {$thisperl .= $Config{_exe} unless $thisperl =~ m/$Config{_exe}$/i;}
497 Don't assume that you can reach the public Internet.
499 Don't assume that there is only one way to get through firewalls
500 to the public Internet.
502 Don't assume that you can reach outside world through any other port
503 than 80, or some web proxy. ftp is blocked by many firewalls.
505 Don't assume that you can send email by connecting to the local SMTP port.
507 Don't assume that you can reach yourself or any node by the name
508 'localhost'. The same goes for '127.0.0.1'. You will have to try both.
510 Don't assume that the host has only one network card, or that it
511 can't bind to many virtual IP addresses.
513 Don't assume a particular network device name.
515 Don't assume a particular set of ioctl()s will work.
517 Don't assume that you can ping hosts and get replies.
519 Don't assume that any particular port (service) will respond.
521 Don't assume that Sys::Hostname (or any other API or command)
522 returns either a fully qualified hostname or a non-qualified hostname:
523 it all depends on how the system had been configured. Also remember
524 things like DHCP and NAT-- the hostname you get back might not be very
527 All the above "don't":s may look daunting, and they are -- but the key
528 is to degrade gracefully if one cannot reach the particular network
529 service one wants. Croaking or hanging do not look very professional.
531 =head2 Interprocess Communication (IPC)
533 In general, don't directly access the system in code meant to be
534 portable. That means, no C<system>, C<exec>, C<fork>, C<pipe>,
535 C<``>, C<qx//>, C<open> with a C<|>, nor any of the other things
536 that makes being a perl hacker worth being.
538 Commands that launch external processes are generally supported on
539 most platforms (though many of them do not support any type of
540 forking). The problem with using them arises from what you invoke
541 them on. External tools are often named differently on different
542 platforms, may not be available in the same location, might accept
543 different arguments, can behave differently, and often present their
544 results in a platform-dependent way. Thus, you should seldom depend
545 on them to produce consistent results. (Then again, if you're calling
546 I<netstat -a>, you probably don't expect it to run on both Unix and CP/M.)
548 One especially common bit of Perl code is opening a pipe to B<sendmail>:
550 open(MAIL, '|/usr/lib/sendmail -t')
551 or die "cannot fork sendmail: $!";
553 This is fine for systems programming when sendmail is known to be
554 available. But it is not fine for many non-Unix systems, and even
555 some Unix systems that may not have sendmail installed. If a portable
556 solution is needed, see the various distributions on CPAN that deal
557 with it. Mail::Mailer and Mail::Send in the MailTools distribution are
558 commonly used, and provide several mailing methods, including mail,
559 sendmail, and direct SMTP (via Net::SMTP) if a mail transfer agent is
560 not available. Mail::Sendmail is a standalone module that provides
561 simple, platform-independent mailing.
563 The Unix System V IPC (C<msg*(), sem*(), shm*()>) is not available
564 even on all Unix platforms.
566 Do not use either the bare result of C<pack("N", 10, 20, 30, 40)> or
567 bare v-strings (such as C<v10.20.30.40>) to represent IPv4 addresses:
568 both forms just pack the four bytes into network order. That this
569 would be equal to the C language C<in_addr> struct (which is what the
570 socket code internally uses) is not guaranteed. To be portable use
571 the routines of the Socket extension, such as C<inet_aton()>,
572 C<inet_ntoa()>, and C<sockaddr_in()>.
574 The rule of thumb for portable code is: Do it all in portable Perl, or
575 use a module (that may internally implement it with platform-specific
576 code, but expose a common interface).
578 =head2 External Subroutines (XS)
580 XS code can usually be made to work with any platform, but dependent
581 libraries, header files, etc., might not be readily available or
582 portable, or the XS code itself might be platform-specific, just as Perl
583 code might be. If the libraries and headers are portable, then it is
584 normally reasonable to make sure the XS code is portable, too.
586 A different type of portability issue arises when writing XS code:
587 availability of a C compiler on the end-user's system. C brings
588 with it its own portability issues, and writing XS code will expose
589 you to some of those. Writing purely in Perl is an easier way to
592 =head2 Standard Modules
594 In general, the standard modules work across platforms. Notable
595 exceptions are the CPAN module (which currently makes connections to external
596 programs that may not be available), platform-specific modules (like
597 ExtUtils::MM_VMS), and DBM modules.
599 There is no one DBM module available on all platforms.
600 SDBM_File and the others are generally available on all Unix and DOSish
601 ports, but not in MacPerl, where only NBDM_File and DB_File are
604 The good news is that at least some DBM module should be available, and
605 AnyDBM_File will use whichever module it can find. Of course, then
606 the code needs to be fairly strict, dropping to the greatest common
607 factor (e.g., not exceeding 1K for each record), so that it will
608 work with any DBM module. See L<AnyDBM_File> for more details.
612 The system's notion of time of day and calendar date is controlled in
613 widely different ways. Don't assume the timezone is stored in C<$ENV{TZ}>,
614 and even if it is, don't assume that you can control the timezone through
615 that variable. Don't assume anything about the three-letter timezone
616 abbreviations (for example that MST would be the Mountain Standard Time,
617 it's been known to stand for Moscow Standard Time). If you need to
618 use timezones, express them in some unambiguous format like the
619 exact number of minutes offset from UTC, or the POSIX timezone
622 Don't assume that the epoch starts at 00:00:00, January 1, 1970,
623 because that is OS- and implementation-specific. It is better to
624 store a date in an unambiguous representation. The ISO 8601 standard
625 defines YYYY-MM-DD as the date format, or YYYY-MM-DDTHH-MM-SS
626 (that's a literal "T" separating the date from the time).
627 Please do use the ISO 8601 instead of making us to guess what
628 date 02/03/04 might be. ISO 8601 even sorts nicely as-is.
629 A text representation (like "1987-12-18") can be easily converted
630 into an OS-specific value using a module like Date::Parse.
631 An array of values, such as those returned by C<localtime>, can be
632 converted to an OS-specific representation using Time::Local.
634 When calculating specific times, such as for tests in time or date modules,
635 it may be appropriate to calculate an offset for the epoch.
638 $offset = Time::Local::timegm(0, 0, 0, 1, 0, 70);
640 The value for C<$offset> in Unix will be C<0>, but in Mac OS will be
641 some large number. C<$offset> can then be added to a Unix time value
642 to get what should be the proper value on any system.
644 On Windows (at least), you shouldn't pass a negative value to C<gmtime> or
647 =head2 Character sets and character encoding
649 Assume very little about character sets.
651 Assume nothing about numerical values (C<ord>, C<chr>) of characters.
652 Do not use explicit code point ranges (like \xHH-\xHH); use for
653 example symbolic character classes like C<[:print:]>.
655 Do not assume that the alphabetic characters are encoded contiguously
656 (in the numeric sense). There may be gaps.
658 Do not assume anything about the ordering of the characters.
659 The lowercase letters may come before or after the uppercase letters;
660 the lowercase and uppercase may be interlaced so that both "a" and "A"
661 come before "b"; the accented and other international characters may
662 be interlaced so that E<auml> comes before "b".
664 =head2 Internationalisation
666 If you may assume POSIX (a rather large assumption), you may read
667 more about the POSIX locale system from L<perllocale>. The locale
668 system at least attempts to make things a little bit more portable,
669 or at least more convenient and native-friendly for non-English
670 users. The system affects character sets and encoding, and date
671 and time formatting--amongst other things.
673 If you really want to be international, you should consider Unicode.
674 See L<perluniintro> and L<perlunicode> for more information.
676 If you want to use non-ASCII bytes (outside the bytes 0x00..0x7f) in
677 the "source code" of your code, to be portable you have to be explicit
678 about what bytes they are. Someone might for example be using your
679 code under a UTF-8 locale, in which case random native bytes might be
680 illegal ("Malformed UTF-8 ...") This means that for example embedding
681 ISO 8859-1 bytes beyond 0x7f into your strings might cause trouble
682 later. If the bytes are native 8-bit bytes, you can use the C<bytes>
683 pragma. If the bytes are in a string (regular expression being a
684 curious string), you can often also use the C<\xHH> notation instead
685 of embedding the bytes as-is. (If you want to write your code in UTF-8,
686 you can use the C<utf8>.) The C<bytes> and C<utf8> pragmata are
687 available since Perl 5.6.0.
689 =head2 System Resources
691 If your code is destined for systems with severely constrained (or
692 missing!) virtual memory systems then you want to be I<especially> mindful
693 of avoiding wasteful constructs such as:
695 # NOTE: this is no longer "bad" in perl5.005
696 for (0..10000000) {} # bad
697 for (my $x = 0; $x <= 10000000; ++$x) {} # good
699 @lines = <VERY_LARGE_FILE>; # bad
701 while (<FILE>) {$file .= $_} # sometimes bad
702 $file = join('', <FILE>); # better
704 The last two constructs may appear unintuitive to most people. The
705 first repeatedly grows a string, whereas the second allocates a
706 large chunk of memory in one go. On some systems, the second is
707 more efficient that the first.
711 Most multi-user platforms provide basic levels of security, usually
712 implemented at the filesystem level. Some, however, do
713 not-- unfortunately. Thus the notion of user id, or "home" directory,
714 or even the state of being logged-in, may be unrecognizable on many
715 platforms. If you write programs that are security-conscious, it
716 is usually best to know what type of system you will be running
717 under so that you can write code explicitly for that platform (or
720 Don't assume the UNIX filesystem access semantics: the operating
721 system or the filesystem may be using some ACL systems, which are
722 richer languages than the usual rwx. Even if the rwx exist,
723 their semantics might be different.
725 (From security viewpoint testing for permissions before attempting to
726 do something is silly anyway: if one tries this, there is potential
727 for race conditions-- someone or something might change the
728 permissions between the permissions check and the actual operation.
729 Just try the operation.)
731 Don't assume the UNIX user and group semantics: especially, don't
732 expect the C<< $< >> and C<< $> >> (or the C<$(> and C<$)>) to work
733 for switching identities (or memberships).
735 Don't assume set-uid and set-gid semantics. (And even if you do,
736 think twice: set-uid and set-gid are a known can of security worms.)
740 For those times when it is necessary to have platform-specific code,
741 consider keeping the platform-specific code in one place, making porting
742 to other platforms easier. Use the Config module and the special
743 variable C<$^O> to differentiate platforms, as described in
746 Be careful in the tests you supply with your module or programs.
747 Module code may be fully portable, but its tests might not be. This
748 often happens when tests spawn off other processes or call external
749 programs to aid in the testing, or when (as noted above) the tests
750 assume certain things about the filesystem and paths. Be careful not
751 to depend on a specific output style for errors, such as when checking
752 C<$!> after a failed system call. Using C<$!> for anything else than
753 displaying it as output is doubtful (though see the Errno module for
754 testing reasonably portably for error value). Some platforms expect
755 a certain output format, and Perl on those platforms may have been
756 adjusted accordingly. Most specifically, don't anchor a regex when
757 testing an error value.
761 Modules uploaded to CPAN are tested by a variety of volunteers on
762 different platforms. These CPAN testers are notified by mail of each
763 new upload, and reply to the list with PASS, FAIL, NA (not applicable to
764 this platform), or UNKNOWN (unknown), along with any relevant notations.
766 The purpose of the testing is twofold: one, to help developers fix any
767 problems in their code that crop up because of lack of testing on other
768 platforms; two, to provide users with information about whether
769 a given module works on a given platform.
777 Mailing list: cpan-testers@perl.org
781 Testing results: http://testers.cpan.org/
787 As of version 5.002, Perl is built with a C<$^O> variable that
788 indicates the operating system it was built on. This was implemented
789 to help speed up code that would otherwise have to C<use Config>
790 and use the value of C<$Config{osname}>. Of course, to get more
791 detailed information about the system, looking into C<%Config> is
792 certainly recommended.
794 C<%Config> cannot always be trusted, however, because it was built
795 at compile time. If perl was built in one place, then transferred
796 elsewhere, some values may be wrong. The values may even have been
797 edited after the fact.
801 Perl works on a bewildering variety of Unix and Unix-like platforms (see
802 e.g. most of the files in the F<hints/> directory in the source code kit).
803 On most of these systems, the value of C<$^O> (hence C<$Config{'osname'}>,
804 too) is determined either by lowercasing and stripping punctuation from the
805 first field of the string returned by typing C<uname -a> (or a similar command)
806 at the shell prompt or by testing the file system for the presence of
807 uniquely named files such as a kernel or header file. Here, for example,
808 are a few of the more popular Unix flavors:
810 uname $^O $Config{'archname'}
811 --------------------------------------------
813 BSD/OS bsdos i386-bsdos
815 dgux dgux AViiON-dgux
816 DYNIX/ptx dynixptx i386-dynixptx
817 FreeBSD freebsd freebsd-i386
818 Haiku haiku BePC-haiku
819 Linux linux arm-linux
820 Linux linux i386-linux
821 Linux linux i586-linux
822 Linux linux ppc-linux
823 HP-UX hpux PA-RISC1.1
825 Mac OS X darwin darwin
826 MachTen PPC machten powerpc-machten
828 NeXT 4 next OPENSTEP-Mach
829 openbsd openbsd i386-openbsd
830 OSF1 dec_osf alpha-dec_osf
831 reliantunix-n svr4 RM400-svr4
832 SCO_SV sco_sv i386-sco_sv
833 SINIX-N svr4 RM400-svr4
834 sn4609 unicos CRAY_C90-unicos
835 sn6521 unicosmk t3e-unicosmk
836 sn9617 unicos CRAY_J90-unicos
837 SunOS solaris sun4-solaris
838 SunOS solaris i86pc-solaris
839 SunOS4 sunos sun4-sunos
841 Because the value of C<$Config{archname}> may depend on the
842 hardware architecture, it can vary more than the value of C<$^O>.
844 =head2 DOS and Derivatives
846 Perl has long been ported to Intel-style microcomputers running under
847 systems like PC-DOS, MS-DOS, OS/2, and most Windows platforms you can
848 bring yourself to mention (except for Windows CE, if you count that).
849 Users familiar with I<COMMAND.COM> or I<CMD.EXE> style shells should
850 be aware that each of these file specifications may have subtle
853 $filespec0 = "c:/foo/bar/file.txt";
854 $filespec1 = "c:\\foo\\bar\\file.txt";
855 $filespec2 = 'c:\foo\bar\file.txt';
856 $filespec3 = 'c:\\foo\\bar\\file.txt';
858 System calls accept either C</> or C<\> as the path separator.
859 However, many command-line utilities of DOS vintage treat C</> as
860 the option prefix, so may get confused by filenames containing C</>.
861 Aside from calling any external programs, C</> will work just fine,
862 and probably better, as it is more consistent with popular usage,
863 and avoids the problem of remembering what to backwhack and what
866 The DOS FAT filesystem can accommodate only "8.3" style filenames. Under
867 the "case-insensitive, but case-preserving" HPFS (OS/2) and NTFS (NT)
868 filesystems you may have to be careful about case returned with functions
869 like C<readdir> or used with functions like C<open> or C<opendir>.
871 DOS also treats several filenames as special, such as AUX, PRN,
872 NUL, CON, COM1, LPT1, LPT2, etc. Unfortunately, sometimes these
873 filenames won't even work if you include an explicit directory
874 prefix. It is best to avoid such filenames, if you want your code
875 to be portable to DOS and its derivatives. It's hard to know what
876 these all are, unfortunately.
878 Users of these operating systems may also wish to make use of
879 scripts such as I<pl2bat.bat> or I<pl2cmd> to
880 put wrappers around your scripts.
882 Newline (C<\n>) is translated as C<\015\012> by STDIO when reading from
883 and writing to files (see L<"Newlines">). C<binmode(FILEHANDLE)>
884 will keep C<\n> translated as C<\012> for that filehandle. Since it is a
885 no-op on other systems, C<binmode> should be used for cross-platform code
886 that deals with binary data. That's assuming you realize in advance
887 that your data is in binary. General-purpose programs should
888 often assume nothing about their data.
890 The C<$^O> variable and the C<$Config{archname}> values for various
891 DOSish perls are as follows:
893 OS $^O $Config{archname} ID Version
894 --------------------------------------------------------
898 Windows 3.1 ? ? 0 3 01
899 Windows 95 MSWin32 MSWin32-x86 1 4 00
900 Windows 98 MSWin32 MSWin32-x86 1 4 10
901 Windows ME MSWin32 MSWin32-x86 1 ?
902 Windows NT MSWin32 MSWin32-x86 2 4 xx
903 Windows NT MSWin32 MSWin32-ALPHA 2 4 xx
904 Windows NT MSWin32 MSWin32-ppc 2 4 xx
905 Windows 2000 MSWin32 MSWin32-x86 2 5 00
906 Windows XP MSWin32 MSWin32-x86 2 5 01
907 Windows 2003 MSWin32 MSWin32-x86 2 5 02
908 Windows CE MSWin32 ? 3
911 The various MSWin32 Perl's can distinguish the OS they are running on
912 via the value of the fifth element of the list returned from
913 Win32::GetOSVersion(). For example:
915 if ($^O eq 'MSWin32') {
916 my @os_version_info = Win32::GetOSVersion();
917 print +('3.1','95','NT')[$os_version_info[4]],"\n";
920 There are also Win32::IsWinNT() and Win32::IsWin95(), try C<perldoc Win32>,
921 and as of libwin32 0.19 (not part of the core Perl distribution)
922 Win32::GetOSName(). The very portable POSIX::uname() will work too:
924 c:\> perl -MPOSIX -we "print join '|', uname"
925 Windows NT|moonru|5.0|Build 2195 (Service Pack 2)|x86
933 The djgpp environment for DOS, http://www.delorie.com/djgpp/
938 The EMX environment for DOS, OS/2, etc. emx@iaehv.nl,
939 http://www.leo.org/pub/comp/os/os2/leo/gnu/emx+gcc/index.html or
940 ftp://hobbes.nmsu.edu/pub/os2/dev/emx/ Also L<perlos2>.
944 Build instructions for Win32 in L<perlwin32>, or under the Cygnus environment
949 The C<Win32::*> modules in L<Win32>.
953 The ActiveState Pages, http://www.activestate.com/
957 The Cygwin environment for Win32; F<README.cygwin> (installed
958 as L<perlcygwin>), http://www.cygwin.com/
962 The U/WIN environment for Win32,
963 http://www.research.att.com/sw/tools/uwin/
967 Build instructions for OS/2, L<perlos2>
973 Any module requiring XS compilation is right out for most people, because
974 MacPerl is built using non-free (and non-cheap!) compilers. Some XS
975 modules that can work with MacPerl are built and distributed in binary
978 Directories are specified as:
980 volume:folder:file for absolute pathnames
981 volume:folder: for absolute pathnames
982 :folder:file for relative pathnames
983 :folder: for relative pathnames
984 :file for relative pathnames
985 file for relative pathnames
987 Files are stored in the directory in alphabetical order. Filenames are
988 limited to 31 characters, and may include any character except for
989 null and C<:>, which is reserved as the path separator.
991 Instead of C<flock>, see C<FSpSetFLock> and C<FSpRstFLock> in the
992 Mac::Files module, or C<chmod(0444, ...)> and C<chmod(0666, ...)>.
994 In the MacPerl application, you can't run a program from the command line;
995 programs that expect C<@ARGV> to be populated can be edited with something
996 like the following, which brings up a dialog box asking for the command
1000 @ARGV = split /\s+/, MacPerl::Ask('Arguments?');
1003 A MacPerl script saved as a "droplet" will populate C<@ARGV> with the full
1004 pathnames of the files dropped onto the script.
1006 Mac users can run programs under a type of command line interface
1007 under MPW (Macintosh Programmer's Workshop, a free development
1008 environment from Apple). MacPerl was first introduced as an MPW
1009 tool, and MPW can be used like a shell:
1011 perl myscript.plx some arguments
1013 ToolServer is another app from Apple that provides access to MPW tools
1014 from MPW and the MacPerl app, which allows MacPerl programs to use
1015 C<system>, backticks, and piped C<open>.
1017 "S<Mac OS>" is the proper name for the operating system, but the value
1018 in C<$^O> is "MacOS". To determine architecture, version, or whether
1019 the application or MPW tool version is running, check:
1021 $is_app = $MacPerl::Version =~ /App/;
1022 $is_tool = $MacPerl::Version =~ /MPW/;
1023 ($version) = $MacPerl::Version =~ /^(\S+)/;
1024 $is_ppc = $MacPerl::Architecture eq 'MacPPC';
1025 $is_68k = $MacPerl::Architecture eq 'Mac68K';
1027 S<Mac OS X>, based on NeXT's OpenStep OS, runs MacPerl natively, under the
1028 "Classic" environment. There is no "Carbon" version of MacPerl to run
1029 under the primary Mac OS X environment. S<Mac OS X> and its Open Source
1030 version, Darwin, both run Unix perl natively.
1038 MacPerl Development, http://dev.macperl.org/ .
1042 The MacPerl Pages, http://www.macperl.com/ .
1046 The MacPerl mailing lists, http://lists.perl.org/ .
1050 MPW, ftp://ftp.apple.com/developer/Tool_Chest/Core_Mac_OS_Tools/
1056 Perl on VMS is discussed in L<perlvms> in the perl distribution.
1058 The official name of VMS as of this writing is OpenVMS.
1060 Perl on VMS can accept either VMS- or Unix-style file
1061 specifications as in either of the following:
1063 $ perl -ne "print if /perl_setup/i" SYS$LOGIN:LOGIN.COM
1064 $ perl -ne "print if /perl_setup/i" /sys$login/login.com
1066 but not a mixture of both as in:
1068 $ perl -ne "print if /perl_setup/i" sys$login:/login.com
1069 Can't open sys$login:/login.com: file specification syntax error
1071 Interacting with Perl from the Digital Command Language (DCL) shell
1072 often requires a different set of quotation marks than Unix shells do.
1075 $ perl -e "print ""Hello, world.\n"""
1078 There are several ways to wrap your perl scripts in DCL F<.COM> files, if
1079 you are so inclined. For example:
1081 $ write sys$output "Hello from DCL!"
1083 $ then perl -x 'f$environment("PROCEDURE")
1084 $ else perl -x - 'p1 'p2 'p3 'p4 'p5 'p6 'p7 'p8
1085 $ deck/dollars="__END__"
1088 print "Hello from Perl!\n";
1093 Do take care with C<$ ASSIGN/nolog/user SYS$COMMAND: SYS$INPUT> if your
1094 perl-in-DCL script expects to do things like C<< $read = <STDIN>; >>.
1096 The VMS operating system has two filesystems, known as ODS-2 and ODS-5.
1098 For ODS-2, filenames are in the format "name.extension;version". The
1099 maximum length for filenames is 39 characters, and the maximum length for
1100 extensions is also 39 characters. Version is a number from 1 to
1101 32767. Valid characters are C</[A-Z0-9$_-]/>.
1103 The ODS-2 filesystem is case-insensitive and does not preserve case.
1104 Perl simulates this by converting all filenames to lowercase internally.
1106 For ODS-5, filenames may have almost any character in them and can include
1107 Unicode characters. Characters that could be misinterpreted by the DCL
1108 shell or file parsing utilities need to be prefixed with the C<^>
1109 character, or replaced with hexadecimal characters prefixed with the
1110 C<^> character. Such prefixing is only needed with the pathnames are
1111 in VMS format in applications. Programs that can accept the UNIX format
1112 of pathnames do not need the escape characters. The maximum length for
1113 filenames is 255 characters. The ODS-5 file system can handle both
1114 a case preserved and a case sensitive mode.
1116 ODS-5 is only available on the OpenVMS for 64 bit platforms.
1118 Support for the extended file specifications is being done as optional
1119 settings to preserve backward compatibility with Perl scripts that
1120 assume the previous VMS limitations.
1122 In general routines on VMS that get a UNIX format file specification
1123 should return it in a UNIX format, and when they get a VMS format
1124 specification they should return a VMS format unless they are documented
1127 For routines that generate return a file specification, VMS allows setting
1128 if the C library which Perl is built on if it will be returned in VMS
1129 format or in UNIX format.
1131 With the ODS-2 file system, there is not much difference in syntax of
1132 filenames without paths for VMS or UNIX. With the extended character
1133 set available with ODS-5 there can be a significant difference.
1135 Because of this, existing Perl scripts written for VMS were sometimes
1136 treating VMS and UNIX filenames interchangeably. Without the extended
1137 character set enabled, this behavior will mostly be maintained for
1138 backwards compatibility.
1140 When extended characters are enabled with ODS-5, the handling of
1141 UNIX formatted file specifications is to that of a UNIX system.
1143 VMS file specifications without extensions have a trailing dot. An
1144 equivalent UNIX file specification should not show the trailing dot.
1146 The result of all of this, is that for VMS, for portable scripts, you
1147 can not depend on Perl to present the filenames in lowercase, to be
1148 case sensitive, and that the filenames could be returned in either
1151 And if a routine returns a file specification, unless it is intended to
1152 convert it, it should return it in the same format as it found it.
1154 C<readdir> by default has traditionally returned lowercased filenames.
1155 When the ODS-5 support is enabled, it will return the exact case of the
1156 filename on the disk.
1158 Files without extensions have a trailing period on them, so doing a
1159 C<readdir> in the default mode with a file named F<A.;5> will
1160 return F<a.> when VMS is (though that file could be opened with
1163 With support for extended file specifications and if C<opendir> was
1164 given a UNIX format directory, a file named F<A.;5> will return F<a>
1165 and optionally in the exact case on the disk. When C<opendir> is given
1166 a VMS format directory, then C<readdir> should return F<a.>, and
1167 again with the optionally the exact case.
1169 RMS had an eight level limit on directory depths from any rooted logical
1170 (allowing 16 levels overall) prior to VMS 7.2, and even with versions of
1171 VMS on VAX up through 7.3. Hence C<PERL_ROOT:[LIB.2.3.4.5.6.7.8]> is a
1172 valid directory specification but C<PERL_ROOT:[LIB.2.3.4.5.6.7.8.9]> is
1173 not. F<Makefile.PL> authors might have to take this into account, but at
1174 least they can refer to the former as C</PERL_ROOT/lib/2/3/4/5/6/7/8/>.
1176 Pumpkings and module integrators can easily see whether files with too many
1177 directory levels have snuck into the core by running the following in the
1178 top-level source directory:
1180 $ perl -ne "$_=~s/\s+.*//; print if scalar(split /\//) > 8;" < MANIFEST
1183 The VMS::Filespec module, which gets installed as part of the build
1184 process on VMS, is a pure Perl module that can easily be installed on
1185 non-VMS platforms and can be helpful for conversions to and from RMS
1186 native formats. It is also now the only way that you should check to
1187 see if VMS is in a case sensitive mode.
1189 What C<\n> represents depends on the type of file opened. It usually
1190 represents C<\012> but it could also be C<\015>, C<\012>, C<\015\012>,
1191 C<\000>, C<\040>, or nothing depending on the file organization and
1192 record format. The VMS::Stdio module provides access to the
1193 special fopen() requirements of files with unusual attributes on VMS.
1195 TCP/IP stacks are optional on VMS, so socket routines might not be
1196 implemented. UDP sockets may not be supported.
1198 The TCP/IP library support for all current versions of VMS is dynamically
1199 loaded if present, so even if the routines are configured, they may
1200 return a status indicating that they are not implemented.
1202 The value of C<$^O> on OpenVMS is "VMS". To determine the architecture
1203 that you are running on without resorting to loading all of C<%Config>
1204 you can examine the content of the C<@INC> array like so:
1206 if (grep(/VMS_AXP/, @INC)) {
1207 print "I'm on Alpha!\n";
1209 } elsif (grep(/VMS_VAX/, @INC)) {
1210 print "I'm on VAX!\n";
1212 } elsif (grep(/VMS_IA64/, @INC)) {
1213 print "I'm on IA64!\n";
1216 print "I'm not so sure about where $^O is...\n";
1219 In general, the significant differences should only be if Perl is running
1220 on VMS_VAX or one of the 64 bit OpenVMS platforms.
1222 On VMS, perl determines the UTC offset from the C<SYS$TIMEZONE_DIFFERENTIAL>
1223 logical name. Although the VMS epoch began at 17-NOV-1858 00:00:00.00,
1224 calls to C<localtime> are adjusted to count offsets from
1225 01-JAN-1970 00:00:00.00, just like Unix.
1233 F<README.vms> (installed as L<README_vms>), L<perlvms>
1237 vmsperl list, vmsperl-subscribe@perl.org
1241 vmsperl on the web, http://www.sidhe.org/vmsperl/index.html
1247 Perl on VOS is discussed in F<README.vos> in the perl distribution
1248 (installed as L<perlvos>). Perl on VOS can accept either VOS- or
1249 Unix-style file specifications as in either of the following:
1251 C<< $ perl -ne "print if /perl_setup/i" >system>notices >>
1252 C<< $ perl -ne "print if /perl_setup/i" /system/notices >>
1254 or even a mixture of both as in:
1256 C<< $ perl -ne "print if /perl_setup/i" >system/notices >>
1258 Even though VOS allows the slash character to appear in object
1259 names, because the VOS port of Perl interprets it as a pathname
1260 delimiting character, VOS files, directories, or links whose names
1261 contain a slash character cannot be processed. Such files must be
1262 renamed before they can be processed by Perl. Note that VOS limits
1263 file names to 32 or fewer characters, file names cannot start with a
1264 C<-> character, or contain any character matching C<< tr/ !%&'()*+;<>?// >>
1266 The value of C<$^O> on VOS is "VOS". To determine the architecture that
1267 you are running on without resorting to loading all of C<%Config> you
1268 can examine the content of the @INC array like so:
1271 print "I'm on a Stratus box!\n";
1273 print "I'm not on a Stratus box!\n";
1283 F<README.vos> (installed as L<perlvos>)
1287 The VOS mailing list.
1289 There is no specific mailing list for Perl on VOS. You can post
1290 comments to the comp.sys.stratus newsgroup, or subscribe to the general
1291 Stratus mailing list. Send a letter with "subscribe Info-Stratus" in
1292 the message body to majordomo@list.stratagy.com.
1296 VOS Perl on the web at http://ftp.stratus.com/pub/vos/posix/posix.html
1300 =head2 EBCDIC Platforms
1302 Recent versions of Perl have been ported to platforms such as OS/400 on
1303 AS/400 minicomputers as well as OS/390, VM/ESA, and BS2000 for S/390
1304 Mainframes. Such computers use EBCDIC character sets internally (usually
1305 Character Code Set ID 0037 for OS/400 and either 1047 or POSIX-BC for S/390
1306 systems). On the mainframe perl currently works under the "Unix system
1307 services for OS/390" (formerly known as OpenEdition), VM/ESA OpenEdition, or
1308 the BS200 POSIX-BC system (BS2000 is supported in perl 5.6 and greater).
1309 See L<perlos390> for details. Note that for OS/400 there is also a port of
1310 Perl 5.8.1/5.9.0 or later to the PASE which is ASCII-based (as opposed to
1311 ILE which is EBCDIC-based), see L<perlos400>.
1313 As of R2.5 of USS for OS/390 and Version 2.3 of VM/ESA these Unix
1314 sub-systems do not support the C<#!> shebang trick for script invocation.
1315 Hence, on OS/390 and VM/ESA perl scripts can be executed with a header
1316 similar to the following simple script:
1319 eval 'exec /usr/local/bin/perl -S $0 ${1+"$@"}'
1321 #!/usr/local/bin/perl # just a comment really
1323 print "Hello from perl!\n";
1325 OS/390 will support the C<#!> shebang trick in release 2.8 and beyond.
1326 Calls to C<system> and backticks can use POSIX shell syntax on all
1329 On the AS/400, if PERL5 is in your library list, you may need
1330 to wrap your perl scripts in a CL procedure to invoke them like so:
1333 CALL PGM(PERL5/PERL) PARM('/QOpenSys/hello.pl')
1336 This will invoke the perl script F<hello.pl> in the root of the
1337 QOpenSys file system. On the AS/400 calls to C<system> or backticks
1340 On these platforms, bear in mind that the EBCDIC character set may have
1341 an effect on what happens with some perl functions (such as C<chr>,
1342 C<pack>, C<print>, C<printf>, C<ord>, C<sort>, C<sprintf>, C<unpack>), as
1343 well as bit-fiddling with ASCII constants using operators like C<^>, C<&>
1344 and C<|>, not to mention dealing with socket interfaces to ASCII computers
1345 (see L<"Newlines">).
1347 Fortunately, most web servers for the mainframe will correctly
1348 translate the C<\n> in the following statement to its ASCII equivalent
1349 (C<\r> is the same under both Unix and OS/390 & VM/ESA):
1351 print "Content-type: text/html\r\n\r\n";
1353 The values of C<$^O> on some of these platforms includes:
1355 uname $^O $Config{'archname'}
1356 --------------------------------------------
1359 POSIX-BC posix-bc BS2000-posix-bc
1362 Some simple tricks for determining if you are running on an EBCDIC
1363 platform could include any of the following (perhaps all):
1365 if ("\t" eq "\05") { print "EBCDIC may be spoken here!\n"; }
1367 if (ord('A') == 193) { print "EBCDIC may be spoken here!\n"; }
1369 if (chr(169) eq 'z') { print "EBCDIC may be spoken here!\n"; }
1371 One thing you may not want to rely on is the EBCDIC encoding
1372 of punctuation characters since these may differ from code page to code
1373 page (and once your module or script is rumoured to work with EBCDIC,
1374 folks will want it to work with all EBCDIC character sets).
1382 L<perlos390>, F<README.os390>, F<perlbs2000>, F<README.vmesa>,
1387 The perl-mvs@perl.org list is for discussion of porting issues as well as
1388 general usage issues for all EBCDIC Perls. Send a message body of
1389 "subscribe perl-mvs" to majordomo@perl.org.
1393 AS/400 Perl information at
1394 http://as400.rochester.ibm.com/
1395 as well as on CPAN in the F<ports/> directory.
1399 =head2 Acorn RISC OS
1401 Because Acorns use ASCII with newlines (C<\n>) in text files as C<\012> like
1402 Unix, and because Unix filename emulation is turned on by default,
1403 most simple scripts will probably work "out of the box". The native
1404 filesystem is modular, and individual filesystems are free to be
1405 case-sensitive or insensitive, and are usually case-preserving. Some
1406 native filesystems have name length limits, which file and directory
1407 names are silently truncated to fit. Scripts should be aware that the
1408 standard filesystem currently has a name length limit of B<10>
1409 characters, with up to 77 items in a directory, but other filesystems
1410 may not impose such limitations.
1412 Native filenames are of the form
1414 Filesystem#Special_Field::DiskName.$.Directory.Directory.File
1418 Special_Field is not usually present, but may contain . and $ .
1419 Filesystem =~ m|[A-Za-z0-9_]|
1420 DsicName =~ m|[A-Za-z0-9_/]|
1421 $ represents the root directory
1422 . is the path separator
1423 @ is the current directory (per filesystem but machine global)
1424 ^ is the parent directory
1425 Directory and File =~ m|[^\0- "\.\$\%\&:\@\\^\|\177]+|
1427 The default filename translation is roughly C<tr|/.|./|;>
1429 Note that C<"ADFS::HardDisk.$.File" ne 'ADFS::HardDisk.$.File'> and that
1430 the second stage of C<$> interpolation in regular expressions will fall
1431 foul of the C<$.> if scripts are not careful.
1433 Logical paths specified by system variables containing comma-separated
1434 search lists are also allowed; hence C<System:Modules> is a valid
1435 filename, and the filesystem will prefix C<Modules> with each section of
1436 C<System$Path> until a name is made that points to an object on disk.
1437 Writing to a new file C<System:Modules> would be allowed only if
1438 C<System$Path> contains a single item list. The filesystem will also
1439 expand system variables in filenames if enclosed in angle brackets, so
1440 C<< <System$Dir>.Modules >> would look for the file
1441 S<C<$ENV{'System$Dir'} . 'Modules'>>. The obvious implication of this is
1442 that B<fully qualified filenames can start with C<< <> >>> and should
1443 be protected when C<open> is used for input.
1445 Because C<.> was in use as a directory separator and filenames could not
1446 be assumed to be unique after 10 characters, Acorn implemented the C
1447 compiler to strip the trailing C<.c> C<.h> C<.s> and C<.o> suffix from
1448 filenames specified in source code and store the respective files in
1449 subdirectories named after the suffix. Hence files are translated:
1452 C:foo.h C:h.foo (logical path variable)
1453 sys/os.h sys.h.os (C compiler groks Unix-speak)
1454 10charname.c c.10charname
1455 10charname.o o.10charname
1456 11charname_.c c.11charname (assuming filesystem truncates at 10)
1458 The Unix emulation library's translation of filenames to native assumes
1459 that this sort of translation is required, and it allows a user-defined list
1460 of known suffixes that it will transpose in this fashion. This may
1461 seem transparent, but consider that with these rules C<foo/bar/baz.h>
1462 and C<foo/bar/h/baz> both map to C<foo.bar.h.baz>, and that C<readdir> and
1463 C<glob> cannot and do not attempt to emulate the reverse mapping. Other
1464 C<.>'s in filenames are translated to C</>.
1466 As implied above, the environment accessed through C<%ENV> is global, and
1467 the convention is that program specific environment variables are of the
1468 form C<Program$Name>. Each filesystem maintains a current directory,
1469 and the current filesystem's current directory is the B<global> current
1470 directory. Consequently, sociable programs don't change the current
1471 directory but rely on full pathnames, and programs (and Makefiles) cannot
1472 assume that they can spawn a child process which can change the current
1473 directory without affecting its parent (and everyone else for that
1476 Because native operating system filehandles are global and are currently
1477 allocated down from 255, with 0 being a reserved value, the Unix emulation
1478 library emulates Unix filehandles. Consequently, you can't rely on
1479 passing C<STDIN>, C<STDOUT>, or C<STDERR> to your children.
1481 The desire of users to express filenames of the form
1482 C<< <Foo$Dir>.Bar >> on the command line unquoted causes problems,
1483 too: C<``> command output capture has to perform a guessing game. It
1484 assumes that a string C<< <[^<>]+\$[^<>]> >> is a
1485 reference to an environment variable, whereas anything else involving
1486 C<< < >> or C<< > >> is redirection, and generally manages to be 99%
1487 right. Of course, the problem remains that scripts cannot rely on any
1488 Unix tools being available, or that any tools found have Unix-like command
1491 Extensions and XS are, in theory, buildable by anyone using free
1492 tools. In practice, many don't, as users of the Acorn platform are
1493 used to binary distributions. MakeMaker does run, but no available
1494 make currently copes with MakeMaker's makefiles; even if and when
1495 this should be fixed, the lack of a Unix-like shell will cause
1496 problems with makefile rules, especially lines of the form C<cd
1497 sdbm && make all>, and anything using quoting.
1499 "S<RISC OS>" is the proper name for the operating system, but the value
1500 in C<$^O> is "riscos" (because we don't like shouting).
1504 Perl has been ported to many platforms that do not fit into any of
1505 the categories listed above. Some, such as AmigaOS, Atari MiNT,
1506 BeOS, HP MPE/iX, QNX, Plan 9, and VOS, have been well-integrated
1507 into the standard Perl source code kit. You may need to see the
1508 F<ports/> directory on CPAN for information, and possibly binaries,
1509 for the likes of: aos, Atari ST, lynxos, riscos, Novell Netware,
1510 Tandem Guardian, I<etc.> (Yes, we know that some of these OSes may
1511 fall under the Unix category, but we are not a standards body.)
1513 Some approximate operating system names and their C<$^O> values
1514 in the "OTHER" category include:
1516 OS $^O $Config{'archname'}
1517 ------------------------------------------
1518 Amiga DOS amigaos m68k-amigos
1520 MPE/iX mpeix PA-RISC1.1
1528 Amiga, F<README.amiga> (installed as L<perlamiga>).
1532 Atari, F<README.mint> and Guido Flohr's web page
1533 http://stud.uni-sb.de/~gufl0000/
1537 Be OS, F<README.beos>
1541 HP 300 MPE/iX, F<README.mpeix> and Mark Bixby's web page
1542 http://www.bixby.org/mark/perlix.html
1546 A free perl5-based PERL.NLM for Novell Netware is available in
1547 precompiled binary and source code form from http://www.novell.com/
1548 as well as from CPAN.
1552 S<Plan 9>, F<README.plan9>
1556 =head1 FUNCTION IMPLEMENTATIONS
1558 Listed below are functions that are either completely unimplemented
1559 or else have been implemented differently on various platforms.
1560 Following each description will be, in parentheses, a list of
1561 platforms that the description applies to.
1563 The list may well be incomplete, or even wrong in some places. When
1564 in doubt, consult the platform-specific README files in the Perl
1565 source distribution, and any other documentation resources accompanying
1568 Be aware, moreover, that even among Unix-ish systems there are variations.
1570 For many functions, you can also query C<%Config>, exported by
1571 default from the Config module. For example, to check whether the
1572 platform has the C<lstat> call, check C<$Config{d_lstat}>. See
1573 L<Config> for a full description of available variables.
1575 =head2 Alphabetical Listing of Perl Functions
1581 C<-r>, C<-w>, and C<-x> have a limited meaning only; directories
1582 and applications are executable, and there are no uid/gid
1583 considerations. C<-o> is not supported. (S<Mac OS>)
1585 C<-w> only inspects the read-only file attribute (FILE_ATTRIBUTE_READONLY),
1586 which determines whether the directory can be deleted, not whether it can
1587 be written to. Directories always have read and write access unless denied
1588 by discretionary access control lists (DACLs). (S<Win32>)
1590 C<-r>, C<-w>, C<-x>, and C<-o> tell whether the file is accessible,
1591 which may not reflect UIC-based file protections. (VMS)
1593 C<-s> returns the size of the data fork, not the total size of data fork
1594 plus resource fork. (S<Mac OS>).
1596 C<-s> by name on an open file will return the space reserved on disk,
1597 rather than the current extent. C<-s> on an open filehandle returns the
1598 current size. (S<RISC OS>)
1600 C<-R>, C<-W>, C<-X>, C<-O> are indistinguishable from C<-r>, C<-w>,
1601 C<-x>, C<-o>. (S<Mac OS>, Win32, VMS, S<RISC OS>)
1603 C<-b>, C<-c>, C<-k>, C<-g>, C<-p>, C<-u>, C<-A> are not implemented.
1606 C<-g>, C<-k>, C<-l>, C<-p>, C<-u>, C<-A> are not particularly meaningful.
1607 (Win32, VMS, S<RISC OS>)
1609 C<-d> is true if passed a device spec without an explicit directory.
1612 C<-T> and C<-B> are implemented, but might misclassify Mac text files
1613 with foreign characters; this is the case will all platforms, but may
1614 affect S<Mac OS> often. (S<Mac OS>)
1616 C<-x> (or C<-X>) determine if a file ends in one of the executable
1617 suffixes. C<-S> is meaningless. (Win32)
1619 C<-x> (or C<-X>) determine if a file has an executable file type.
1624 Due to issues with various CPUs, math libraries, compilers, and standards,
1625 results for C<atan2()> may vary depending on any combination of the above.
1626 Perl attempts to conform to the Open Group/IEEE standards for the results
1627 returned from C<atan2()>, but cannot force the issue if the system Perl is
1628 run on does not allow it. (Tru64, HP-UX 10.20)
1630 The current version of the standards for C<atan2()> is available at
1631 L<http://www.opengroup.org/onlinepubs/009695399/functions/atan2.html>.
1635 Meaningless. (S<Mac OS>, S<RISC OS>)
1637 Reopens file and restores pointer; if function fails, underlying
1638 filehandle may be closed, or pointer may be in a different position.
1641 The value returned by C<tell> may be affected after the call, and
1642 the filehandle may be flushed. (Win32)
1646 Only limited meaning. Disabling/enabling write permission is mapped to
1647 locking/unlocking the file. (S<Mac OS>)
1649 Only good for changing "owner" read-write access, "group", and "other"
1650 bits are meaningless. (Win32)
1652 Only good for changing "owner" and "other" read-write access. (S<RISC OS>)
1654 Access permissions are mapped onto VOS access-control list changes. (VOS)
1656 The actual permissions set depend on the value of the C<CYGWIN>
1657 in the SYSTEM environment settings. (Cygwin)
1661 Not implemented. (S<Mac OS>, Win32, S<Plan 9>, S<RISC OS>)
1663 Does nothing, but won't fail. (Win32)
1665 A little funky, because VOS's notion of ownership is a little funky (VOS).
1669 Not implemented. (S<Mac OS>, Win32, VMS, S<Plan 9>, S<RISC OS>, VOS, VM/ESA)
1673 May not be available if library or source was not provided when building
1678 Not implemented. (VMS, S<Plan 9>, VOS)
1682 Not implemented. (VMS, S<Plan 9>, VOS)
1686 Not useful. (S<Mac OS>, S<RISC OS>)
1688 Not supported. (Cygwin, Win32)
1690 Invokes VMS debugger. (VMS)
1694 Not implemented. (S<Mac OS>)
1696 Implemented via Spawn. (VM/ESA)
1698 Does not automatically flush output handles on some platforms.
1699 (SunOS, Solaris, HP-UX)
1703 Emulates UNIX exit() (which considers C<exit 1> to indicate an error) by
1704 mapping the C<1> to SS$_ABORT (C<44>). This behavior may be overridden
1705 with the pragma C<use vmsish 'exit'>. As with the CRTL's exit()
1706 function, C<exit 0> is also mapped to an exit status of SS$_NORMAL
1707 (C<1>); this mapping cannot be overridden. Any other argument to exit()
1708 is used directly as Perl's exit status. On VMS, unless the future
1709 POSIX_EXIT mode is enabled, the exit code should always be a valid
1710 VMS exit code and not a generic number. When the POSIX_EXIT mode is
1711 enabled, a generic number will be encoded in a method compatible with
1712 the C library _POSIX_EXIT macro so that it can be decoded by other
1713 programs, particularly ones written in C, like the GNV package. (VMS)
1717 Not implemented. (Win32)
1718 Some functions available based on the version of VMS. (VMS)
1722 Not implemented (S<Mac OS>, VMS, S<RISC OS>, VOS).
1724 Available only on Windows NT (not on Windows 95). (Win32)
1728 Not implemented. (S<Mac OS>, AmigaOS, S<RISC OS>, VM/ESA, VMS)
1730 Emulated using multiple interpreters. See L<perlfork>. (Win32)
1732 Does not automatically flush output handles on some platforms.
1733 (SunOS, Solaris, HP-UX)
1737 Not implemented. (S<Mac OS>, S<RISC OS>)
1741 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>)
1745 Not implemented. (S<Mac OS>, Win32, S<RISC OS>)
1749 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS, VM/ESA)
1753 Not implemented. (S<Mac OS>, Win32)
1755 Not useful. (S<RISC OS>)
1759 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>)
1763 Not implemented. (S<Mac OS>, Win32, S<Plan 9>)
1767 Not implemented. (S<Mac OS>, Win32)
1769 Not useful. (S<RISC OS>)
1773 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>)
1777 Not implemented. (S<Mac OS>, Win32, S<Plan 9>)
1779 =item getprotobynumber
1781 Not implemented. (S<Mac OS>)
1785 Not implemented. (S<Mac OS>)
1789 Not implemented. (S<Mac OS>, Win32, VM/ESA)
1793 Not implemented. (S<Mac OS>, Win32, VMS, VM/ESA)
1797 C<gethostbyname('localhost')> does not work everywhere: you may have
1798 to use C<gethostbyname('127.0.0.1')>. (S<Mac OS>, S<Irix 5>)
1802 Not implemented. (S<Mac OS>, Win32)
1806 Not implemented. (S<Mac OS>, Win32, S<Plan 9>)
1810 Not implemented. (S<Mac OS>, Win32, S<Plan 9>)
1814 Not implemented. (Win32, S<Plan 9>)
1818 Not implemented. (S<Mac OS>, Win32, S<Plan 9>, S<RISC OS>)
1822 Not implemented. (S<Mac OS>, Win32, S<Plan 9>, S<RISC OS>)
1826 Not implemented. (S<Mac OS>, Win32, S<Plan 9>, S<RISC OS>)
1830 Not implemented. (S<Plan 9>, Win32, S<RISC OS>)
1834 Not implemented. (S<Mac OS>, MPE/iX, VM/ESA, Win32)
1838 Not implemented. (S<Mac OS>, MPE/iX, S<RISC OS>, VM/ESA, VMS, Win32)
1842 Not implemented. (S<Mac OS>, Win32)
1846 Not implemented. (S<Mac OS>, Win32, S<Plan 9>)
1850 Not implemented. (S<Mac OS>, Win32, S<Plan 9>)
1854 Not implemented. (S<Plan 9>, Win32)
1856 =item getsockopt SOCKET,LEVEL,OPTNAME
1858 Not implemented. (S<Plan 9>)
1862 This operator is implemented via the File::Glob extension on most
1863 platforms. See L<File::Glob> for portability information.
1867 Same portability caveats as L<localtime>.
1869 =item ioctl FILEHANDLE,FUNCTION,SCALAR
1871 Not implemented. (VMS)
1873 Available only for socket handles, and it does what the ioctlsocket() call
1874 in the Winsock API does. (Win32)
1876 Available only for socket handles. (S<RISC OS>)
1880 C<kill(0, LIST)> is implemented for the sake of taint checking;
1881 use with other signals is unimplemented. (S<Mac OS>)
1883 Not implemented, hence not useful for taint checking. (S<RISC OS>)
1885 C<kill()> doesn't have the semantics of C<raise()>, i.e. it doesn't send
1886 a signal to the identified process like it does on Unix platforms.
1887 Instead C<kill($sig, $pid)> terminates the process identified by $pid,
1888 and makes it exit immediately with exit status $sig. As in Unix, if
1889 $sig is 0 and the specified process exists, it returns true without
1890 actually terminating it. (Win32)
1892 C<kill(-9, $pid)> will terminate the process specified by $pid and
1893 recursively all child processes owned by it. This is different from
1894 the Unix semantics, where the signal will be delivered to all
1895 processes in the same process group as the process specified by
1898 Is not supported for process identification number of 0 or negative
1903 Not implemented. (S<Mac OS>, MPE/iX, S<RISC OS>)
1905 Link count not updated because hard links are not quite that hard
1906 (They are sort of half-way between hard and soft links). (AmigaOS)
1908 Hard links are implemented on Win32 under NTFS only. They are
1909 natively supported on Windows 2000 and later. On Windows NT they
1910 are implemented using the Windows POSIX subsystem support and the
1911 Perl process will need Administrator or Backup Operator privileges
1912 to create hard links.
1914 Available on 64 bit OpenVMS 8.2 and later. (VMS)
1918 Because Perl currently relies on the native standard C localtime()
1919 function, it is only safe to use times between 0 and (2**31)-1. Times
1920 outside this range may result in unexpected behavior depending on your
1921 operating system's implementation of localtime().
1925 Not implemented. (S<RISC OS>)
1927 Return values (especially for device and inode) may be bogus. (Win32)
1937 Not implemented. (S<Mac OS>, Win32, VMS, S<Plan 9>, S<RISC OS>, VOS)
1941 The C<|> variants are supported only if ToolServer is installed.
1944 open to C<|-> and C<-|> are unsupported. (S<Mac OS>, Win32, S<RISC OS>)
1946 Opening a process does not automatically flush output handles on some
1947 platforms. (SunOS, Solaris, HP-UX)
1951 Very limited functionality. (MiNT)
1955 Not implemented. (Win32, VMS, S<RISC OS>)
1959 Can't move directories between directories on different logical volumes. (Win32)
1963 Only implemented on sockets. (Win32, VMS)
1965 Only reliable on sockets. (S<RISC OS>)
1967 Note that the C<select FILEHANDLE> form is generally portable.
1975 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS)
1979 Not implemented. (S<Mac OS>, MPE/iX, VMS, Win32, S<RISC OS>, VOS)
1983 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS)
1987 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS)
1991 Not implemented. (S<Mac OS>, MPE/iX, Win32, S<RISC OS>, VOS)
1995 Not implemented. (S<Plan 9>)
2005 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS)
2009 A relatively recent addition to socket functions, may not
2010 be implemented even in UNIX platforms.
2014 Not implemented. (S<RISC OS>, VOS, VM/ESA)
2016 Available on 64 bit OpenVMS 8.2 and later. (VMS)
2020 Platforms that do not have rdev, blksize, or blocks will return these
2021 as '', so numeric comparison or manipulation of these fields may cause
2022 'not numeric' warnings.
2024 mtime and atime are the same thing, and ctime is creation time instead of
2025 inode change time. (S<Mac OS>).
2027 ctime not supported on UFS (S<Mac OS X>).
2029 ctime is creation time instead of inode change time (Win32).
2031 device and inode are not meaningful. (Win32)
2033 device and inode are not necessarily reliable. (VMS)
2035 mtime, atime and ctime all return the last modification time. Device and
2036 inode are not necessarily reliable. (S<RISC OS>)
2038 dev, rdev, blksize, and blocks are not available. inode is not
2039 meaningful and will differ between stat calls on the same file. (os2)
2041 some versions of cygwin when doing a stat("foo") and if not finding it
2042 may then attempt to stat("foo.exe") (Cygwin)
2044 On Win32 stat() needs to open the file to determine the link count
2045 and update attributes that may have been changed through hard links.
2046 Setting ${^WIN32_SLOPPY_STAT} to a true value speeds up stat() by
2047 not performing this operation. (Win32)
2051 Not implemented. (Win32, S<RISC OS>)
2053 Implemented on 64 bit VMS 8.3. VMS requires the symbolic link to be in Unix
2054 syntax if it is intended to resolve to a valid path.
2058 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS, VM/ESA)
2062 The traditional "0", "1", and "2" MODEs are implemented with different
2063 numeric values on some systems. The flags exported by C<Fcntl>
2064 (O_RDONLY, O_WRONLY, O_RDWR) should work everywhere though. (S<Mac
2065 OS>, OS/390, VM/ESA)
2069 Only implemented if ToolServer is installed. (S<Mac OS>)
2071 As an optimization, may not call the command shell specified in
2072 C<$ENV{PERL5SHELL}>. C<system(1, @args)> spawns an external
2073 process and immediately returns its process designator, without
2074 waiting for it to terminate. Return value may be used subsequently
2075 in C<wait> or C<waitpid>. Failure to spawn() a subprocess is indicated
2076 by setting $? to "255 << 8". C<$?> is set in a way compatible with
2077 Unix (i.e. the exitstatus of the subprocess is obtained by "$? >> 8",
2078 as described in the documentation). (Win32)
2080 There is no shell to process metacharacters, and the native standard is
2081 to pass a command line terminated by "\n" "\r" or "\0" to the spawned
2082 program. Redirection such as C<< > foo >> is performed (if at all) by
2083 the run time library of the spawned program. C<system> I<list> will call
2084 the Unix emulation library's C<exec> emulation, which attempts to provide
2085 emulation of the stdin, stdout, stderr in force in the parent, providing
2086 the child program uses a compatible version of the emulation library.
2087 I<scalar> will call the native command line direct and no such emulation
2088 of a child Unix program will exists. Mileage B<will> vary. (S<RISC OS>)
2090 Far from being POSIX compliant. Because there may be no underlying
2091 /bin/sh tries to work around the problem by forking and execing the
2092 first token in its argument string. Handles basic redirection
2093 ("<" or ">") on its own behalf. (MiNT)
2095 Does not automatically flush output handles on some platforms.
2096 (SunOS, Solaris, HP-UX)
2098 The return value is POSIX-like (shifted up by 8 bits), which only allows
2099 room for a made-up value derived from the severity bits of the native
2100 32-bit condition code (unless overridden by C<use vmsish 'status'>).
2101 If the native condition code is one that has a POSIX value encoded, the
2102 POSIX value will be decoded to extract the expected exit value.
2103 For more details see L<perlvms/$?>. (VMS)
2107 Only the first entry returned is nonzero. (S<Mac OS>)
2109 "cumulative" times will be bogus. On anything other than Windows NT
2110 or Windows 2000, "system" time will be bogus, and "user" time is
2111 actually the time returned by the clock() function in the C runtime
2114 Not useful. (S<RISC OS>)
2118 Not implemented. (Older versions of VMS)
2120 Truncation to same-or-shorter lengths only. (VOS)
2122 If a FILEHANDLE is supplied, it must be writable and opened in append
2123 mode (i.e., use C<<< open(FH, '>>filename') >>>
2124 or C<sysopen(FH,...,O_APPEND|O_RDWR)>. If a filename is supplied, it
2125 should not be held open elsewhere. (Win32)
2129 Returns undef where unavailable, as of version 5.005.
2131 C<umask> works but the correct permissions are set only when the file
2132 is finally closed. (AmigaOS)
2136 Only the modification time is updated. (S<BeOS>, S<Mac OS>, VMS, S<RISC OS>)
2138 May not behave as expected. Behavior depends on the C runtime
2139 library's implementation of utime(), and the filesystem being
2140 used. The FAT filesystem typically does not support an "access
2141 time" field, and it may limit timestamps to a granularity of
2142 two seconds. (Win32)
2148 Not implemented. (S<Mac OS>)
2150 Can only be applied to process handles returned for processes spawned
2151 using C<system(1, ...)> or pseudo processes created with C<fork()>. (Win32)
2153 Not useful. (S<RISC OS>)
2158 =head1 Supported Platforms
2160 As of July 2002 (the Perl release 5.8.0), the following platforms are
2161 able to build Perl from the standard source code distribution
2162 available at http://www.cpan.org/src/index.html
2173 HI-UXMPP (Hitachi) (5.8.0 worked but we didn't know it)
2183 ReliantUNIX (formerly SINIX)
2185 OpenVMS (formerly VMS)
2186 Open UNIX (Unixware) (since Perl 5.8.1/5.9.0)
2188 OS/400 (using the PASE) (since Perl 5.8.1/5.9.0)
2190 POSIX-BC (formerly BS2000)
2195 Tru64 UNIX (formerly DEC OSF/1, Digital UNIX)
2200 Win95/98/ME/2K/XP 2)
2202 z/OS (formerly OS/390)
2205 1) in DOS mode either the DOS or OS/2 ports can be used
2206 2) compilers: Borland, MinGW (GCC), VC6
2208 The following platforms worked with the previous releases (5.6 and
2209 5.7), but we did not manage either to fix or to test these in time
2210 for the 5.8.0 release. There is a very good chance that many of these
2211 will work fine with the 5.8.0.
2224 Known to be broken for 5.8.0 (but 5.6.1 and 5.7.2 can be used):
2228 The following platforms have been known to build Perl from source in
2229 the past (5.005_03 and earlier), but we haven't been able to verify
2230 their status for the current release, either because the
2231 hardware/software platforms are rare or because we don't have an
2232 active champion on these platforms--or both. They used to work,
2233 though, so go ahead and try compiling them, and let perlbug@perl.org
2267 The following platforms have their own source code distributions and
2268 binaries available via http://www.cpan.org/ports/
2272 OS/400 (ILE) 5.005_02
2273 Tandem Guardian 5.004
2275 The following platforms have only binaries available via
2276 http://www.cpan.org/ports/index.html :
2280 Acorn RISCOS 5.005_02
2284 Although we do suggest that you always build your own Perl from
2285 the source code, both for maximal configurability and for security,
2286 in case you are in a hurry you can check
2287 http://www.cpan.org/ports/index.html for binary distributions.
2291 L<perlaix>, L<perlamiga>, L<perlapollo>, L<perlbeos>, L<perlbs2000>,
2292 L<perlce>, L<perlcygwin>, L<perldgux>, L<perldos>, L<perlepoc>,
2293 L<perlebcdic>, L<perlfreebsd>, L<perlhurd>, L<perlhpux>, L<perlirix>,
2294 L<perlmachten>, L<perlmacos>, L<perlmacosx>, L<perlmint>, L<perlmpeix>,
2295 L<perlnetware>, L<perlos2>, L<perlos390>, L<perlos400>,
2296 L<perlplan9>, L<perlqnx>, L<perlsolaris>, L<perltru64>,
2297 L<perlunicode>, L<perlvmesa>, L<perlvms>, L<perlvos>,
2298 L<perlwin32>, and L<Win32>.
2300 =head1 AUTHORS / CONTRIBUTORS
2302 Abigail <abigail@foad.org>,
2303 Charles Bailey <bailey@newman.upenn.edu>,
2304 Graham Barr <gbarr@pobox.com>,
2305 Tom Christiansen <tchrist@perl.com>,
2306 Nicholas Clark <nick@ccl4.org>,
2307 Thomas Dorner <Thomas.Dorner@start.de>,
2308 Andy Dougherty <doughera@lafayette.edu>,
2309 Dominic Dunlop <domo@computer.org>,
2310 Neale Ferguson <neale@vma.tabnsw.com.au>,
2311 David J. Fiander <davidf@mks.com>,
2312 Paul Green <Paul.Green@stratus.com>,
2313 M.J.T. Guy <mjtg@cam.ac.uk>,
2314 Jarkko Hietaniemi <jhi@iki.fi>,
2315 Luther Huffman <lutherh@stratcom.com>,
2316 Nick Ing-Simmons <nick@ing-simmons.net>,
2317 Andreas J. KE<ouml>nig <a.koenig@mind.de>,
2318 Markus Laker <mlaker@contax.co.uk>,
2319 Andrew M. Langmead <aml@world.std.com>,
2320 Larry Moore <ljmoore@freespace.net>,
2321 Paul Moore <Paul.Moore@uk.origin-it.com>,
2322 Chris Nandor <pudge@pobox.com>,
2323 Matthias Neeracher <neeracher@mac.com>,
2324 Philip Newton <pne@cpan.org>,
2325 Gary Ng <71564.1743@CompuServe.COM>,
2326 Tom Phoenix <rootbeer@teleport.com>,
2327 AndrE<eacute> Pirard <A.Pirard@ulg.ac.be>,
2328 Peter Prymmer <pvhp@forte.com>,
2329 Hugo van der Sanden <hv@crypt0.demon.co.uk>,
2330 Gurusamy Sarathy <gsar@activestate.com>,
2331 Paul J. Schinder <schinder@pobox.com>,
2332 Michael G Schwern <schwern@pobox.com>,
2333 Dan Sugalski <dan@sidhe.org>,
2334 Nathan Torkington <gnat@frii.com>.
2335 John Malmberg <wb8tyw@qsl.net>