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 (<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 A common cause of unportable programs is the misuse of chop() to trim
107 You can get away with this on Unix and MacOS (they have a single
108 character end-of-line), but the same program will break under DOSish
109 perls because you're only chop()ing half the end-of-line. Instead,
110 chomp() should be used to trim newlines. The Dunce::Files module can
111 help audit your code for misuses of chop().
113 When dealing with binary files (or text files in binary mode) be sure
114 to explicitly set $/ to the appropriate value for your file format
115 before using chomp().
117 Because of the "text" mode translation, DOSish perls have limitations
118 in using C<seek> and C<tell> on a file accessed in "text" mode.
119 Stick to C<seek>-ing to locations you got from C<tell> (and no
120 others), and you are usually free to use C<seek> and C<tell> even
121 in "text" mode. Using C<seek> or C<tell> or other file operations
122 may be non-portable. If you use C<binmode> on a file, however, you
123 can usually C<seek> and C<tell> with arbitrary values in safety.
125 A common misconception in socket programming is that C<\n> eq C<\012>
126 everywhere. When using protocols such as common Internet protocols,
127 C<\012> and C<\015> are called for specifically, and the values of
128 the logical C<\n> and C<\r> (carriage return) are not reliable.
130 print SOCKET "Hi there, client!\r\n"; # WRONG
131 print SOCKET "Hi there, client!\015\012"; # RIGHT
133 However, using C<\015\012> (or C<\cM\cJ>, or C<\x0D\x0A>) can be tedious
134 and unsightly, as well as confusing to those maintaining the code. As
135 such, the Socket module supplies the Right Thing for those who want it.
137 use Socket qw(:DEFAULT :crlf);
138 print SOCKET "Hi there, client!$CRLF" # RIGHT
140 When reading from a socket, remember that the default input record
141 separator C<$/> is C<\n>, but robust socket code will recognize as
142 either C<\012> or C<\015\012> as end of line:
148 Because both CRLF and LF end in LF, the input record separator can
149 be set to LF and any CR stripped later. Better to write:
151 use Socket qw(:DEFAULT :crlf);
152 local($/) = LF; # not needed if $/ is already \012
155 s/$CR?$LF/\n/; # not sure if socket uses LF or CRLF, OK
156 # s/\015?\012/\n/; # same thing
159 This example is preferred over the previous one--even for Unix
160 platforms--because now any C<\015>'s (C<\cM>'s) are stripped out
161 (and there was much rejoicing).
163 Similarly, functions that return text data--such as a function that
164 fetches a web page--should sometimes translate newlines before
165 returning the data, if they've not yet been translated to the local
166 newline representation. A single line of code will often suffice:
168 $data =~ s/\015?\012/\n/g;
171 Some of this may be confusing. Here's a handy reference to the ASCII CR
172 and LF characters. You can print it out and stick it in your wallet.
174 LF == \012 == \x0A == \cJ == ASCII 10
175 CR == \015 == \x0D == \cM == ASCII 13
178 ---------------------------
181 \n * | LF | CRLF | CR |
182 \r * | CR | CR | LF |
183 ---------------------------
186 The Unix column assumes that you are not accessing a serial line
187 (like a tty) in canonical mode. If you are, then CR on input becomes
188 "\n", and "\n" on output becomes CRLF.
190 These are just the most common definitions of C<\n> and C<\r> in Perl.
191 There may well be others.
193 =head2 Numbers endianness and Width
195 Different CPUs store integers and floating point numbers in different
196 orders (called I<endianness>) and widths (32-bit and 64-bit being the
197 most common today). This affects your programs when they attempt to transfer
198 numbers in binary format from one CPU architecture to another,
199 usually either "live" via network connection, or by storing the
200 numbers to secondary storage such as a disk file or tape.
202 Conflicting storage orders make utter mess out of the numbers. If a
203 little-endian host (Intel, VAX) stores 0x12345678 (305419896 in
204 decimal), a big-endian host (Motorola, Sparc, PA) reads it as
205 0x78563412 (2018915346 in decimal). Alpha and MIPS can be either:
206 Digital/Compaq used/uses them in little-endian mode; SGI/Cray uses
207 them in big-endian mode. To avoid this problem in network (socket)
208 connections use the C<pack> and C<unpack> formats C<n> and C<N>, the
209 "network" orders. These are guaranteed to be portable.
211 You can explore the endianness of your platform by unpacking a
212 data structure packed in native format such as:
214 print unpack("h*", pack("s2", 1, 2)), "\n";
215 # '10002000' on e.g. Intel x86 or Alpha 21064 in little-endian mode
216 # '00100020' on e.g. Motorola 68040
218 If you need to distinguish between endian architectures you could use
219 either of the variables set like so:
221 $is_big_endian = unpack("h*", pack("s", 1)) =~ /01/;
222 $is_little_endian = unpack("h*", pack("s", 1)) =~ /^1/;
224 Differing widths can cause truncation even between platforms of equal
225 endianness. The platform of shorter width loses the upper parts of the
226 number. There is no good solution for this problem except to avoid
227 transferring or storing raw binary numbers.
229 One can circumnavigate both these problems in two ways. Either
230 transfer and store numbers always in text format, instead of raw
231 binary, or else consider using modules like Data::Dumper (included in
232 the standard distribution as of Perl 5.005) and Storable (included as
233 of perl 5.8). Keeping all data as text significantly simplifies matters.
235 =head2 Files and Filesystems
237 Most platforms these days structure files in a hierarchical fashion.
238 So, it is reasonably safe to assume that all platforms support the
239 notion of a "path" to uniquely identify a file on the system. How
240 that path is really written, though, differs considerably.
242 Although similar, file path specifications differ between Unix,
243 Windows, S<Mac OS>, OS/2, VMS, VOS, S<RISC OS>, and probably others.
244 Unix, for example, is one of the few OSes that has the elegant idea
245 of a single root directory.
247 DOS, OS/2, VMS, VOS, and Windows can work similarly to Unix with C</>
248 as path separator, or in their own idiosyncratic ways (such as having
249 several root directories and various "unrooted" device files such NIL:
252 S<Mac OS> uses C<:> as a path separator instead of C</>.
254 The filesystem may support neither hard links (C<link>) nor
255 symbolic links (C<symlink>, C<readlink>, C<lstat>).
257 The filesystem may support neither access timestamp nor change
258 timestamp (meaning that about the only portable timestamp is the
259 modification timestamp), or one second granularity of any timestamps
260 (e.g. the FAT filesystem limits the time granularity to two seconds).
262 VOS perl can emulate Unix filenames with C</> as path separator. The
263 native pathname characters greater-than, less-than, number-sign, and
264 percent-sign are always accepted.
266 S<RISC OS> perl can emulate Unix filenames with C</> as path
267 separator, or go native and use C<.> for path separator and C<:> to
268 signal filesystems and disk names.
270 Don't assume UNIX filesystem access semantics: that read, write,
271 and execute are all the permissions there are, and even if they exist,
272 that their semantics (for example what do r, w, and x mean on
273 a directory) are the UNIX ones. The various UNIX/POSIX compatibility
274 layers usually try to make interfaces like chmod() work, but sometimes
275 there simply is no good mapping.
277 If all this is intimidating, have no (well, maybe only a little)
278 fear. There are modules that can help. The File::Spec modules
279 provide methods to do the Right Thing on whatever platform happens
280 to be running the program.
282 use File::Spec::Functions;
283 chdir(updir()); # go up one directory
284 $file = catfile(curdir(), 'temp', 'file.txt');
285 # on Unix and Win32, './temp/file.txt'
286 # on Mac OS, ':temp:file.txt'
287 # on VMS, '[.temp]file.txt'
289 File::Spec is available in the standard distribution as of version
290 5.004_05. File::Spec::Functions is only in File::Spec 0.7 and later,
291 and some versions of perl come with version 0.6. If File::Spec
292 is not updated to 0.7 or later, you must use the object-oriented
293 interface from File::Spec (or upgrade File::Spec).
295 In general, production code should not have file paths hardcoded.
296 Making them user-supplied or read from a configuration file is
297 better, keeping in mind that file path syntax varies on different
300 This is especially noticeable in scripts like Makefiles and test suites,
301 which often assume C</> as a path separator for subdirectories.
303 Also of use is File::Basename from the standard distribution, which
304 splits a pathname into pieces (base filename, full path to directory,
307 Even when on a single platform (if you can call Unix a single platform),
308 remember not to count on the existence or the contents of particular
309 system-specific files or directories, like F</etc/passwd>,
310 F</etc/sendmail.conf>, F</etc/resolv.conf>, or even F</tmp/>. For
311 example, F</etc/passwd> may exist but not contain the encrypted
312 passwords, because the system is using some form of enhanced security.
313 Or it may not contain all the accounts, because the system is using NIS.
314 If code does need to rely on such a file, include a description of the
315 file and its format in the code's documentation, then make it easy for
316 the user to override the default location of the file.
318 Don't assume a text file will end with a newline. They should,
321 Do not have two files of the same name with different case, like
322 F<test.pl> and F<Test.pl>, as many platforms have case-insensitive
323 filenames. Also, try not to have non-word characters (except for C<.>)
324 in the names, and keep them to the 8.3 convention, for maximum
325 portability, onerous a burden though this may appear.
327 Likewise, when using the AutoSplit module, try to keep your functions to
328 8.3 naming and case-insensitive conventions; or, at the least,
329 make it so the resulting files have a unique (case-insensitively)
332 Whitespace in filenames is tolerated on most systems, but not all.
333 Many systems (DOS, VMS) cannot have more than one C<.> in their filenames.
335 Don't assume C<< > >> won't be the first character of a filename.
336 Always use C<< < >> explicitly to open a file for reading,
337 unless you want the user to be able to specify a pipe open.
339 open(FILE, "< $existing_file") or die $!;
341 If filenames might use strange characters, it is safest to open it
342 with C<sysopen> instead of C<open>. C<open> is magic and can
343 translate characters like C<< > >>, C<< < >>, and C<|>, which may
344 be the wrong thing to do. (Sometimes, though, it's the right thing.)
346 =head2 System Interaction
348 Not all platforms provide a command line. These are usually platforms
349 that rely primarily on a Graphical User Interface (GUI) for user
350 interaction. A program requiring a command line interface might
351 not work everywhere. This is probably for the user of the program
352 to deal with, so don't stay up late worrying about it.
354 Some platforms can't delete or rename files held open by the system.
355 Remember to C<close> files when you are done with them. Don't
356 C<unlink> or C<rename> an open file. Don't C<tie> or C<open> a
357 file already tied or opened; C<untie> or C<close> it first.
359 Don't open the same file more than once at a time for writing, as some
360 operating systems put mandatory locks on such files.
362 Don't assume that write/modify permission on a directory gives the
363 right to add or delete files/directories in that directory. That is
364 filesystem specific: in some filesystems you need write/modify
365 permission also (or even just) in the file/directory itself. In some
366 filesystems (AFS, DFS) the permission to add/delete directory entries
367 is a completely separate permission.
369 Don't assume that a single C<unlink> completely gets rid of the file:
370 some filesystems (most notably the ones in VMS) have versioned
371 filesystems, and unlink() removes only the most recent one (it doesn't
372 remove all the versions because by default the native tools on those
373 platforms remove just the most recent version, too). The portable
374 idiom to remove all the versions of a file is
376 1 while unlink "file";
378 This will terminate if the file is undeleteable for some reason
379 (protected, not there, and so on).
381 Don't count on a specific environment variable existing in C<%ENV>.
382 Don't count on C<%ENV> entries being case-sensitive, or even
383 case-preserving. Don't try to clear %ENV by saying C<%ENV = ();>, or,
384 if you really have to, make it conditional on C<$^O ne 'VMS'> since in
385 VMS the C<%ENV> table is much more than a per-process key-value string
388 Don't count on signals or C<%SIG> for anything.
390 Don't count on filename globbing. Use C<opendir>, C<readdir>, and
393 Don't count on per-program environment variables, or per-program current
396 Don't count on specific values of C<$!>.
398 =head2 Interprocess Communication (IPC)
400 In general, don't directly access the system in code meant to be
401 portable. That means, no C<system>, C<exec>, C<fork>, C<pipe>,
402 C<``>, C<qx//>, C<open> with a C<|>, nor any of the other things
403 that makes being a perl hacker worth being.
405 Commands that launch external processes are generally supported on
406 most platforms (though many of them do not support any type of
407 forking). The problem with using them arises from what you invoke
408 them on. External tools are often named differently on different
409 platforms, may not be available in the same location, might accept
410 different arguments, can behave differently, and often present their
411 results in a platform-dependent way. Thus, you should seldom depend
412 on them to produce consistent results. (Then again, if you're calling
413 I<netstat -a>, you probably don't expect it to run on both Unix and CP/M.)
415 One especially common bit of Perl code is opening a pipe to B<sendmail>:
417 open(MAIL, '|/usr/lib/sendmail -t')
418 or die "cannot fork sendmail: $!";
420 This is fine for systems programming when sendmail is known to be
421 available. But it is not fine for many non-Unix systems, and even
422 some Unix systems that may not have sendmail installed. If a portable
423 solution is needed, see the various distributions on CPAN that deal
424 with it. Mail::Mailer and Mail::Send in the MailTools distribution are
425 commonly used, and provide several mailing methods, including mail,
426 sendmail, and direct SMTP (via Net::SMTP) if a mail transfer agent is
427 not available. Mail::Sendmail is a standalone module that provides
428 simple, platform-independent mailing.
430 The Unix System V IPC (C<msg*(), sem*(), shm*()>) is not available
431 even on all Unix platforms.
433 Do not use either the bare result of C<pack("N", 10, 20, 30, 40)> or
434 bare v-strings (such as C<v10.20.30.40>) to represent IPv4 addresses:
435 both forms just pack the four bytes into network order. That this
436 would be equal to the C language C<in_addr> struct (which is what the
437 socket code internally uses) is not guaranteed. To be portable use
438 the routines of the Socket extension, such as C<inet_aton()>,
439 C<inet_ntoa()>, and C<sockaddr_in()>.
441 The rule of thumb for portable code is: Do it all in portable Perl, or
442 use a module (that may internally implement it with platform-specific
443 code, but expose a common interface).
445 =head2 External Subroutines (XS)
447 XS code can usually be made to work with any platform, but dependent
448 libraries, header files, etc., might not be readily available or
449 portable, or the XS code itself might be platform-specific, just as Perl
450 code might be. If the libraries and headers are portable, then it is
451 normally reasonable to make sure the XS code is portable, too.
453 A different type of portability issue arises when writing XS code:
454 availability of a C compiler on the end-user's system. C brings
455 with it its own portability issues, and writing XS code will expose
456 you to some of those. Writing purely in Perl is an easier way to
459 =head2 Standard Modules
461 In general, the standard modules work across platforms. Notable
462 exceptions are the CPAN module (which currently makes connections to external
463 programs that may not be available), platform-specific modules (like
464 ExtUtils::MM_VMS), and DBM modules.
466 There is no one DBM module available on all platforms.
467 SDBM_File and the others are generally available on all Unix and DOSish
468 ports, but not in MacPerl, where only NBDM_File and DB_File are
471 The good news is that at least some DBM module should be available, and
472 AnyDBM_File will use whichever module it can find. Of course, then
473 the code needs to be fairly strict, dropping to the greatest common
474 factor (e.g., not exceeding 1K for each record), so that it will
475 work with any DBM module. See L<AnyDBM_File> for more details.
479 The system's notion of time of day and calendar date is controlled in
480 widely different ways. Don't assume the timezone is stored in C<$ENV{TZ}>,
481 and even if it is, don't assume that you can control the timezone through
484 Don't assume that the epoch starts at 00:00:00, January 1, 1970,
485 because that is OS- and implementation-specific. It is better to store a date
486 in an unambiguous representation. The ISO-8601 standard defines
487 "YYYY-MM-DD" as the date format. A text representation (like "1987-12-18")
488 can be easily converted into an OS-specific value using a module like
489 Date::Parse. An array of values, such as those returned by
490 C<localtime>, can be converted to an OS-specific representation using
493 When calculating specific times, such as for tests in time or date modules,
494 it may be appropriate to calculate an offset for the epoch.
497 $offset = Time::Local::timegm(0, 0, 0, 1, 0, 70);
499 The value for C<$offset> in Unix will be C<0>, but in Mac OS will be
500 some large number. C<$offset> can then be added to a Unix time value
501 to get what should be the proper value on any system.
503 =head2 Character sets and character encoding
505 Assume little about character sets. Assume nothing about
506 numerical values (C<ord>, C<chr>) of characters. Do not
507 assume that the alphabetic characters are encoded contiguously (in
508 the numeric sense). Do not assume anything about the ordering of the
509 characters. The lowercase letters may come before or after the
510 uppercase letters; the lowercase and uppercase may be interlaced so
511 that both `a' and `A' come before `b'; the accented and other
512 international characters may be interlaced so that E<auml> comes
515 =head2 Internationalisation
517 If you may assume POSIX (a rather large assumption), you may read
518 more about the POSIX locale system from L<perllocale>. The locale
519 system at least attempts to make things a little bit more portable,
520 or at least more convenient and native-friendly for non-English
521 users. The system affects character sets and encoding, and date
522 and time formatting--amongst other things.
524 =head2 System Resources
526 If your code is destined for systems with severely constrained (or
527 missing!) virtual memory systems then you want to be I<especially> mindful
528 of avoiding wasteful constructs such as:
530 # NOTE: this is no longer "bad" in perl5.005
531 for (0..10000000) {} # bad
532 for (my $x = 0; $x <= 10000000; ++$x) {} # good
534 @lines = <VERY_LARGE_FILE>; # bad
536 while (<FILE>) {$file .= $_} # sometimes bad
537 $file = join('', <FILE>); # better
539 The last two constructs may appear unintuitive to most people. The
540 first repeatedly grows a string, whereas the second allocates a
541 large chunk of memory in one go. On some systems, the second is
542 more efficient that the first.
546 Most multi-user platforms provide basic levels of security, usually
547 implemented at the filesystem level. Some, however, do
548 not-- unfortunately. Thus the notion of user id, or "home" directory,
549 or even the state of being logged-in, may be unrecognizable on many
550 platforms. If you write programs that are security-conscious, it
551 is usually best to know what type of system you will be running
552 under so that you can write code explicitly for that platform (or
555 Don't assume the UNIX filesystem access semantics: the operating
556 system or the filesystem may be using some ACL systems, which are
557 richer languages than the usual rwx. Even if the rwx exist,
558 their semantics might be different.
560 (From security viewpoint testing for permissions before attempting to
561 do something is silly anyway: if one tries this, there is potential
562 for race conditions-- someone or something might change the
563 permissions between the permissions check and the actual operation.
564 Just try the operation.)
566 Don't assume the UNIX user and group semantics: especially, don't
567 expect the C<< $< >> and C<< $> >> (or the C<$(> and C<$)>) to work
568 for switching identities (or memberships).
570 Don't assume set-uid and set-gid semantics. (And even if you do,
571 think twice: set-uid and set-gid are a known can of security worms.)
575 For those times when it is necessary to have platform-specific code,
576 consider keeping the platform-specific code in one place, making porting
577 to other platforms easier. Use the Config module and the special
578 variable C<$^O> to differentiate platforms, as described in
581 Be careful in the tests you supply with your module or programs.
582 Module code may be fully portable, but its tests might not be. This
583 often happens when tests spawn off other processes or call external
584 programs to aid in the testing, or when (as noted above) the tests
585 assume certain things about the filesystem and paths. Be careful
586 not to depend on a specific output style for errors, such as when
587 checking C<$!> after a system call. Some platforms expect a certain
588 output format, and perl on those platforms may have been adjusted
589 accordingly. Most specifically, don't anchor a regex when testing
594 Modules uploaded to CPAN are tested by a variety of volunteers on
595 different platforms. These CPAN testers are notified by mail of each
596 new upload, and reply to the list with PASS, FAIL, NA (not applicable to
597 this platform), or UNKNOWN (unknown), along with any relevant notations.
599 The purpose of the testing is twofold: one, to help developers fix any
600 problems in their code that crop up because of lack of testing on other
601 platforms; two, to provide users with information about whether
602 a given module works on a given platform.
606 =item Mailing list: cpan-testers@perl.org
608 =item Testing results: http://testers.cpan.org/
614 As of version 5.002, Perl is built with a C<$^O> variable that
615 indicates the operating system it was built on. This was implemented
616 to help speed up code that would otherwise have to C<use Config>
617 and use the value of C<$Config{osname}>. Of course, to get more
618 detailed information about the system, looking into C<%Config> is
619 certainly recommended.
621 C<%Config> cannot always be trusted, however, because it was built
622 at compile time. If perl was built in one place, then transferred
623 elsewhere, some values may be wrong. The values may even have been
624 edited after the fact.
628 Perl works on a bewildering variety of Unix and Unix-like platforms (see
629 e.g. most of the files in the F<hints/> directory in the source code kit).
630 On most of these systems, the value of C<$^O> (hence C<$Config{'osname'}>,
631 too) is determined either by lowercasing and stripping punctuation from the
632 first field of the string returned by typing C<uname -a> (or a similar command)
633 at the shell prompt or by testing the file system for the presence of
634 uniquely named files such as a kernel or header file. Here, for example,
635 are a few of the more popular Unix flavors:
637 uname $^O $Config{'archname'}
638 --------------------------------------------
640 BSD/OS bsdos i386-bsdos
641 dgux dgux AViiON-dgux
642 DYNIX/ptx dynixptx i386-dynixptx
643 FreeBSD freebsd freebsd-i386
644 Linux linux arm-linux
645 Linux linux i386-linux
646 Linux linux i586-linux
647 Linux linux ppc-linux
648 HP-UX hpux PA-RISC1.1
650 Mac OS X darwin darwin
651 MachTen PPC machten powerpc-machten
653 NeXT 4 next OPENSTEP-Mach
654 openbsd openbsd i386-openbsd
655 OSF1 dec_osf alpha-dec_osf
656 reliantunix-n svr4 RM400-svr4
657 SCO_SV sco_sv i386-sco_sv
658 SINIX-N svr4 RM400-svr4
659 sn4609 unicos CRAY_C90-unicos
660 sn6521 unicosmk t3e-unicosmk
661 sn9617 unicos CRAY_J90-unicos
662 SunOS solaris sun4-solaris
663 SunOS solaris i86pc-solaris
664 SunOS4 sunos sun4-sunos
666 Because the value of C<$Config{archname}> may depend on the
667 hardware architecture, it can vary more than the value of C<$^O>.
669 =head2 DOS and Derivatives
671 Perl has long been ported to Intel-style microcomputers running under
672 systems like PC-DOS, MS-DOS, OS/2, and most Windows platforms you can
673 bring yourself to mention (except for Windows CE, if you count that).
674 Users familiar with I<COMMAND.COM> or I<CMD.EXE> style shells should
675 be aware that each of these file specifications may have subtle
678 $filespec0 = "c:/foo/bar/file.txt";
679 $filespec1 = "c:\\foo\\bar\\file.txt";
680 $filespec2 = 'c:\foo\bar\file.txt';
681 $filespec3 = 'c:\\foo\\bar\\file.txt';
683 System calls accept either C</> or C<\> as the path separator.
684 However, many command-line utilities of DOS vintage treat C</> as
685 the option prefix, so may get confused by filenames containing C</>.
686 Aside from calling any external programs, C</> will work just fine,
687 and probably better, as it is more consistent with popular usage,
688 and avoids the problem of remembering what to backwhack and what
691 The DOS FAT filesystem can accommodate only "8.3" style filenames. Under
692 the "case-insensitive, but case-preserving" HPFS (OS/2) and NTFS (NT)
693 filesystems you may have to be careful about case returned with functions
694 like C<readdir> or used with functions like C<open> or C<opendir>.
696 DOS also treats several filenames as special, such as AUX, PRN,
697 NUL, CON, COM1, LPT1, LPT2, etc. Unfortunately, sometimes these
698 filenames won't even work if you include an explicit directory
699 prefix. It is best to avoid such filenames, if you want your code
700 to be portable to DOS and its derivatives. It's hard to know what
701 these all are, unfortunately.
703 Users of these operating systems may also wish to make use of
704 scripts such as I<pl2bat.bat> or I<pl2cmd> to
705 put wrappers around your scripts.
707 Newline (C<\n>) is translated as C<\015\012> by STDIO when reading from
708 and writing to files (see L<"Newlines">). C<binmode(FILEHANDLE)>
709 will keep C<\n> translated as C<\012> for that filehandle. Since it is a
710 no-op on other systems, C<binmode> should be used for cross-platform code
711 that deals with binary data. That's assuming you realize in advance
712 that your data is in binary. General-purpose programs should
713 often assume nothing about their data.
715 The C<$^O> variable and the C<$Config{archname}> values for various
716 DOSish perls are as follows:
718 OS $^O $Config{archname} ID Version
719 --------------------------------------------------------
723 Windows 3.1 ? ? 0 3 01
724 Windows 95 MSWin32 MSWin32-x86 1 4 00
725 Windows 98 MSWin32 MSWin32-x86 1 4 10
726 Windows ME MSWin32 MSWin32-x86 1 ?
727 Windows NT MSWin32 MSWin32-x86 2 4 xx
728 Windows NT MSWin32 MSWin32-ALPHA 2 4 xx
729 Windows NT MSWin32 MSWin32-ppc 2 4 xx
730 Windows 2000 MSWin32 MSWin32-x86 2 5 xx
731 Windows XP MSWin32 MSWin32-x86 2 ?
732 Windows CE MSWin32 ? 3
735 The various MSWin32 Perl's can distinguish the OS they are running on
736 via the value of the fifth element of the list returned from
737 Win32::GetOSVersion(). For example:
739 if ($^O eq 'MSWin32') {
740 my @os_version_info = Win32::GetOSVersion();
741 print +('3.1','95','NT')[$os_version_info[4]],"\n";
750 The djgpp environment for DOS, http://www.delorie.com/djgpp/
755 The EMX environment for DOS, OS/2, etc. emx@iaehv.nl,
756 http://www.leo.org/pub/comp/os/os2/leo/gnu/emx+gcc/index.html or
757 ftp://hobbes.nmsu.edu/pub/os2/dev/emx. Also L<perlos2>.
761 Build instructions for Win32 in L<perlwin32>, or under the Cygnus environment
766 The C<Win32::*> modules in L<Win32>.
770 The ActiveState Pages, http://www.activestate.com/
774 The Cygwin environment for Win32; F<README.cygwin> (installed
775 as L<perlcygwin>), http://www.cygwin.com/
779 The U/WIN environment for Win32,
780 http://www.research.att.com/sw/tools/uwin/
784 Build instructions for OS/2, L<perlos2>
790 Any module requiring XS compilation is right out for most people, because
791 MacPerl is built using non-free (and non-cheap!) compilers. Some XS
792 modules that can work with MacPerl are built and distributed in binary
795 Directories are specified as:
797 volume:folder:file for absolute pathnames
798 volume:folder: for absolute pathnames
799 :folder:file for relative pathnames
800 :folder: for relative pathnames
801 :file for relative pathnames
802 file for relative pathnames
804 Files are stored in the directory in alphabetical order. Filenames are
805 limited to 31 characters, and may include any character except for
806 null and C<:>, which is reserved as the path separator.
808 Instead of C<flock>, see C<FSpSetFLock> and C<FSpRstFLock> in the
809 Mac::Files module, or C<chmod(0444, ...)> and C<chmod(0666, ...)>.
811 In the MacPerl application, you can't run a program from the command line;
812 programs that expect C<@ARGV> to be populated can be edited with something
813 like the following, which brings up a dialog box asking for the command
817 @ARGV = split /\s+/, MacPerl::Ask('Arguments?');
820 A MacPerl script saved as a "droplet" will populate C<@ARGV> with the full
821 pathnames of the files dropped onto the script.
823 Mac users can run programs under a type of command line interface
824 under MPW (Macintosh Programmer's Workshop, a free development
825 environment from Apple). MacPerl was first introduced as an MPW
826 tool, and MPW can be used like a shell:
828 perl myscript.plx some arguments
830 ToolServer is another app from Apple that provides access to MPW tools
831 from MPW and the MacPerl app, which allows MacPerl programs to use
832 C<system>, backticks, and piped C<open>.
834 "S<Mac OS>" is the proper name for the operating system, but the value
835 in C<$^O> is "MacOS". To determine architecture, version, or whether
836 the application or MPW tool version is running, check:
838 $is_app = $MacPerl::Version =~ /App/;
839 $is_tool = $MacPerl::Version =~ /MPW/;
840 ($version) = $MacPerl::Version =~ /^(\S+)/;
841 $is_ppc = $MacPerl::Architecture eq 'MacPPC';
842 $is_68k = $MacPerl::Architecture eq 'Mac68K';
844 S<Mac OS X>, based on NeXT's OpenStep OS, runs MacPerl natively, under the
845 "Classic" environment. There is no "Carbon" version of MacPerl to run
846 under the primary Mac OS X environment. S<Mac OS X> and its Open Source
847 version, Darwin, both run Unix perl natively.
855 MacPerl Development, http://dev.macperl.org/ .
859 The MacPerl Pages, http://www.macperl.com/ .
863 The MacPerl mailing lists, http://lists.perl.org/ .
869 Perl on VMS is discussed in L<perlvms> in the perl distribution.
870 Perl on VMS can accept either VMS- or Unix-style file
871 specifications as in either of the following:
873 $ perl -ne "print if /perl_setup/i" SYS$LOGIN:LOGIN.COM
874 $ perl -ne "print if /perl_setup/i" /sys$login/login.com
876 but not a mixture of both as in:
878 $ perl -ne "print if /perl_setup/i" sys$login:/login.com
879 Can't open sys$login:/login.com: file specification syntax error
881 Interacting with Perl from the Digital Command Language (DCL) shell
882 often requires a different set of quotation marks than Unix shells do.
885 $ perl -e "print ""Hello, world.\n"""
888 There are several ways to wrap your perl scripts in DCL F<.COM> files, if
889 you are so inclined. For example:
891 $ write sys$output "Hello from DCL!"
893 $ then perl -x 'f$environment("PROCEDURE")
894 $ else perl -x - 'p1 'p2 'p3 'p4 'p5 'p6 'p7 'p8
895 $ deck/dollars="__END__"
898 print "Hello from Perl!\n";
903 Do take care with C<$ ASSIGN/nolog/user SYS$COMMAND: SYS$INPUT> if your
904 perl-in-DCL script expects to do things like C<< $read = <STDIN>; >>.
906 Filenames are in the format "name.extension;version". The maximum
907 length for filenames is 39 characters, and the maximum length for
908 extensions is also 39 characters. Version is a number from 1 to
909 32767. Valid characters are C</[A-Z0-9$_-]/>.
911 VMS's RMS filesystem is case-insensitive and does not preserve case.
912 C<readdir> returns lowercased filenames, but specifying a file for
913 opening remains case-insensitive. Files without extensions have a
914 trailing period on them, so doing a C<readdir> with a file named F<A.;5>
915 will return F<a.> (though that file could be opened with
918 RMS had an eight level limit on directory depths from any rooted logical
919 (allowing 16 levels overall) prior to VMS 7.2. Hence
920 C<PERL_ROOT:[LIB.2.3.4.5.6.7.8]> is a valid directory specification but
921 C<PERL_ROOT:[LIB.2.3.4.5.6.7.8.9]> is not. F<Makefile.PL> authors might
922 have to take this into account, but at least they can refer to the former
923 as C</PERL_ROOT/lib/2/3/4/5/6/7/8/>.
925 The VMS::Filespec module, which gets installed as part of the build
926 process on VMS, is a pure Perl module that can easily be installed on
927 non-VMS platforms and can be helpful for conversions to and from RMS
930 What C<\n> represents depends on the type of file opened. It usually
931 represents C<\012> but it could also be C<\015>, C<\012>, C<\015\012>,
932 C<\000>, C<\040>, or nothing depending on the file organiztion and
933 record format. The VMS::Stdio module provides access to the
934 special fopen() requirements of files with unusual attributes on VMS.
936 TCP/IP stacks are optional on VMS, so socket routines might not be
937 implemented. UDP sockets may not be supported.
939 The value of C<$^O> on OpenVMS is "VMS". To determine the architecture
940 that you are running on without resorting to loading all of C<%Config>
941 you can examine the content of the C<@INC> array like so:
943 if (grep(/VMS_AXP/, @INC)) {
944 print "I'm on Alpha!\n";
946 } elsif (grep(/VMS_VAX/, @INC)) {
947 print "I'm on VAX!\n";
950 print "I'm not so sure about where $^O is...\n";
953 On VMS, perl determines the UTC offset from the C<SYS$TIMEZONE_DIFFERENTIAL>
954 logical name. Although the VMS epoch began at 17-NOV-1858 00:00:00.00,
955 calls to C<localtime> are adjusted to count offsets from
956 01-JAN-1970 00:00:00.00, just like Unix.
964 F<README.vms> (installed as L<README_vms>), L<perlvms>
968 vmsperl list, majordomo@perl.org
970 (Put the words C<subscribe vmsperl> in message body.)
974 vmsperl on the web, http://www.sidhe.org/vmsperl/index.html
980 Perl on VOS is discussed in F<README.vos> in the perl distribution
981 (installed as L<perlvos>). Perl on VOS can accept either VOS- or
982 Unix-style file specifications as in either of the following:
984 $ perl -ne "print if /perl_setup/i" >system>notices
985 $ perl -ne "print if /perl_setup/i" /system/notices
987 or even a mixture of both as in:
989 $ perl -ne "print if /perl_setup/i" >system/notices
991 Even though VOS allows the slash character to appear in object
992 names, because the VOS port of Perl interprets it as a pathname
993 delimiting character, VOS files, directories, or links whose names
994 contain a slash character cannot be processed. Such files must be
995 renamed before they can be processed by Perl. Note that VOS limits
996 file names to 32 or fewer characters.
998 See F<README.vos> for restrictions that apply when Perl is built
999 with the alpha version of VOS POSIX.1 support.
1001 Perl on VOS is built without any extensions and does not support
1004 The value of C<$^O> on VOS is "VOS". To determine the architecture that
1005 you are running on without resorting to loading all of C<%Config> you
1006 can examine the content of the @INC array like so:
1009 print "I'm on a Stratus box!\n";
1011 print "I'm not on a Stratus box!\n";
1015 if (grep(/860/, @INC)) {
1016 print "This box is a Stratus XA/R!\n";
1018 } elsif (grep(/7100/, @INC)) {
1019 print "This box is a Stratus HP 7100 or 8xxx!\n";
1021 } elsif (grep(/8000/, @INC)) {
1022 print "This box is a Stratus HP 8xxx!\n";
1025 print "This box is a Stratus 68K!\n";
1038 The VOS mailing list.
1040 There is no specific mailing list for Perl on VOS. You can post
1041 comments to the comp.sys.stratus newsgroup, or subscribe to the general
1042 Stratus mailing list. Send a letter with "Subscribe Info-Stratus" in
1043 the message body to majordomo@list.stratagy.com.
1047 VOS Perl on the web at http://ftp.stratus.com/pub/vos/vos.html
1051 =head2 EBCDIC Platforms
1053 Recent versions of Perl have been ported to platforms such as OS/400 on
1054 AS/400 minicomputers as well as OS/390, VM/ESA, and BS2000 for S/390
1055 Mainframes. Such computers use EBCDIC character sets internally (usually
1056 Character Code Set ID 0037 for OS/400 and either 1047 or POSIX-BC for S/390
1057 systems). On the mainframe perl currently works under the "Unix system
1058 services for OS/390" (formerly known as OpenEdition), VM/ESA OpenEdition, or
1059 the BS200 POSIX-BC system (BS2000 is supported in perl 5.6 and greater).
1060 See L<perlos390> for details.
1062 As of R2.5 of USS for OS/390 and Version 2.3 of VM/ESA these Unix
1063 sub-systems do not support the C<#!> shebang trick for script invocation.
1064 Hence, on OS/390 and VM/ESA perl scripts can be executed with a header
1065 similar to the following simple script:
1068 eval 'exec /usr/local/bin/perl -S $0 ${1+"$@"}'
1070 #!/usr/local/bin/perl # just a comment really
1072 print "Hello from perl!\n";
1074 OS/390 will support the C<#!> shebang trick in release 2.8 and beyond.
1075 Calls to C<system> and backticks can use POSIX shell syntax on all
1078 On the AS/400, if PERL5 is in your library list, you may need
1079 to wrap your perl scripts in a CL procedure to invoke them like so:
1082 CALL PGM(PERL5/PERL) PARM('/QOpenSys/hello.pl')
1085 This will invoke the perl script F<hello.pl> in the root of the
1086 QOpenSys file system. On the AS/400 calls to C<system> or backticks
1089 On these platforms, bear in mind that the EBCDIC character set may have
1090 an effect on what happens with some perl functions (such as C<chr>,
1091 C<pack>, C<print>, C<printf>, C<ord>, C<sort>, C<sprintf>, C<unpack>), as
1092 well as bit-fiddling with ASCII constants using operators like C<^>, C<&>
1093 and C<|>, not to mention dealing with socket interfaces to ASCII computers
1094 (see L<"Newlines">).
1096 Fortunately, most web servers for the mainframe will correctly
1097 translate the C<\n> in the following statement to its ASCII equivalent
1098 (C<\r> is the same under both Unix and OS/390 & VM/ESA):
1100 print "Content-type: text/html\r\n\r\n";
1102 The values of C<$^O> on some of these platforms includes:
1104 uname $^O $Config{'archname'}
1105 --------------------------------------------
1108 POSIX-BC posix-bc BS2000-posix-bc
1111 Some simple tricks for determining if you are running on an EBCDIC
1112 platform could include any of the following (perhaps all):
1114 if ("\t" eq "\05") { print "EBCDIC may be spoken here!\n"; }
1116 if (ord('A') == 193) { print "EBCDIC may be spoken here!\n"; }
1118 if (chr(169) eq 'z') { print "EBCDIC may be spoken here!\n"; }
1120 One thing you may not want to rely on is the EBCDIC encoding
1121 of punctuation characters since these may differ from code page to code
1122 page (and once your module or script is rumoured to work with EBCDIC,
1123 folks will want it to work with all EBCDIC character sets).
1133 L<perlos390>, F<README.os390>, F<perlbs2000>, F<README.vmesa>,
1138 The perl-mvs@perl.org list is for discussion of porting issues as well as
1139 general usage issues for all EBCDIC Perls. Send a message body of
1140 "subscribe perl-mvs" to majordomo@perl.org.
1144 AS/400 Perl information at
1145 http://as400.rochester.ibm.com/
1146 as well as on CPAN in the F<ports/> directory.
1150 =head2 Acorn RISC OS
1152 Because Acorns use ASCII with newlines (C<\n>) in text files as C<\012> like
1153 Unix, and because Unix filename emulation is turned on by default,
1154 most simple scripts will probably work "out of the box". The native
1155 filesystem is modular, and individual filesystems are free to be
1156 case-sensitive or insensitive, and are usually case-preserving. Some
1157 native filesystems have name length limits, which file and directory
1158 names are silently truncated to fit. Scripts should be aware that the
1159 standard filesystem currently has a name length limit of B<10>
1160 characters, with up to 77 items in a directory, but other filesystems
1161 may not impose such limitations.
1163 Native filenames are of the form
1165 Filesystem#Special_Field::DiskName.$.Directory.Directory.File
1169 Special_Field is not usually present, but may contain . and $ .
1170 Filesystem =~ m|[A-Za-z0-9_]|
1171 DsicName =~ m|[A-Za-z0-9_/]|
1172 $ represents the root directory
1173 . is the path separator
1174 @ is the current directory (per filesystem but machine global)
1175 ^ is the parent directory
1176 Directory and File =~ m|[^\0- "\.\$\%\&:\@\\^\|\177]+|
1178 The default filename translation is roughly C<tr|/.|./|;>
1180 Note that C<"ADFS::HardDisk.$.File" ne 'ADFS::HardDisk.$.File'> and that
1181 the second stage of C<$> interpolation in regular expressions will fall
1182 foul of the C<$.> if scripts are not careful.
1184 Logical paths specified by system variables containing comma-separated
1185 search lists are also allowed; hence C<System:Modules> is a valid
1186 filename, and the filesystem will prefix C<Modules> with each section of
1187 C<System$Path> until a name is made that points to an object on disk.
1188 Writing to a new file C<System:Modules> would be allowed only if
1189 C<System$Path> contains a single item list. The filesystem will also
1190 expand system variables in filenames if enclosed in angle brackets, so
1191 C<< <System$Dir>.Modules >> would look for the file
1192 S<C<$ENV{'System$Dir'} . 'Modules'>>. The obvious implication of this is
1193 that B<fully qualified filenames can start with C<< <> >>> and should
1194 be protected when C<open> is used for input.
1196 Because C<.> was in use as a directory separator and filenames could not
1197 be assumed to be unique after 10 characters, Acorn implemented the C
1198 compiler to strip the trailing C<.c> C<.h> C<.s> and C<.o> suffix from
1199 filenames specified in source code and store the respective files in
1200 subdirectories named after the suffix. Hence files are translated:
1203 C:foo.h C:h.foo (logical path variable)
1204 sys/os.h sys.h.os (C compiler groks Unix-speak)
1205 10charname.c c.10charname
1206 10charname.o o.10charname
1207 11charname_.c c.11charname (assuming filesystem truncates at 10)
1209 The Unix emulation library's translation of filenames to native assumes
1210 that this sort of translation is required, and it allows a user-defined list
1211 of known suffixes that it will transpose in this fashion. This may
1212 seem transparent, but consider that with these rules C<foo/bar/baz.h>
1213 and C<foo/bar/h/baz> both map to C<foo.bar.h.baz>, and that C<readdir> and
1214 C<glob> cannot and do not attempt to emulate the reverse mapping. Other
1215 C<.>'s in filenames are translated to C</>.
1217 As implied above, the environment accessed through C<%ENV> is global, and
1218 the convention is that program specific environment variables are of the
1219 form C<Program$Name>. Each filesystem maintains a current directory,
1220 and the current filesystem's current directory is the B<global> current
1221 directory. Consequently, sociable programs don't change the current
1222 directory but rely on full pathnames, and programs (and Makefiles) cannot
1223 assume that they can spawn a child process which can change the current
1224 directory without affecting its parent (and everyone else for that
1227 Because native operating system filehandles are global and are currently
1228 allocated down from 255, with 0 being a reserved value, the Unix emulation
1229 library emulates Unix filehandles. Consequently, you can't rely on
1230 passing C<STDIN>, C<STDOUT>, or C<STDERR> to your children.
1232 The desire of users to express filenames of the form
1233 C<< <Foo$Dir>.Bar >> on the command line unquoted causes problems,
1234 too: C<``> command output capture has to perform a guessing game. It
1235 assumes that a string C<< <[^<>]+\$[^<>]> >> is a
1236 reference to an environment variable, whereas anything else involving
1237 C<< < >> or C<< > >> is redirection, and generally manages to be 99%
1238 right. Of course, the problem remains that scripts cannot rely on any
1239 Unix tools being available, or that any tools found have Unix-like command
1242 Extensions and XS are, in theory, buildable by anyone using free
1243 tools. In practice, many don't, as users of the Acorn platform are
1244 used to binary distributions. MakeMaker does run, but no available
1245 make currently copes with MakeMaker's makefiles; even if and when
1246 this should be fixed, the lack of a Unix-like shell will cause
1247 problems with makefile rules, especially lines of the form C<cd
1248 sdbm && make all>, and anything using quoting.
1250 "S<RISC OS>" is the proper name for the operating system, but the value
1251 in C<$^O> is "riscos" (because we don't like shouting).
1255 Perl has been ported to many platforms that do not fit into any of
1256 the categories listed above. Some, such as AmigaOS, Atari MiNT,
1257 BeOS, HP MPE/iX, QNX, Plan 9, and VOS, have been well-integrated
1258 into the standard Perl source code kit. You may need to see the
1259 F<ports/> directory on CPAN for information, and possibly binaries,
1260 for the likes of: aos, Atari ST, lynxos, riscos, Novell Netware,
1261 Tandem Guardian, I<etc.> (Yes, we know that some of these OSes may
1262 fall under the Unix category, but we are not a standards body.)
1264 Some approximate operating system names and their C<$^O> values
1265 in the "OTHER" category include:
1267 OS $^O $Config{'archname'}
1268 ------------------------------------------
1269 Amiga DOS amigaos m68k-amigos
1270 MPE/iX mpeix PA-RISC1.1
1278 Amiga, F<README.amiga> (installed as L<perlamiga>).
1282 Atari, F<README.mint> and Guido Flohr's web page
1283 http://stud.uni-sb.de/~gufl0000/
1287 Be OS, F<README.beos>
1291 HP 300 MPE/iX, F<README.mpeix> and Mark Bixby's web page
1292 http://www.bixby.org/mark/perlix.html
1296 A free perl5-based PERL.NLM for Novell Netware is available in
1297 precompiled binary and source code form from http://www.novell.com/
1298 as well as from CPAN.
1302 Plan 9, F<README.plan9>
1306 =head1 FUNCTION IMPLEMENTATIONS
1308 Listed below are functions that are either completely unimplemented
1309 or else have been implemented differently on various platforms.
1310 Following each description will be, in parentheses, a list of
1311 platforms that the description applies to.
1313 The list may well be incomplete, or even wrong in some places. When
1314 in doubt, consult the platform-specific README files in the Perl
1315 source distribution, and any other documentation resources accompanying
1318 Be aware, moreover, that even among Unix-ish systems there are variations.
1320 For many functions, you can also query C<%Config>, exported by
1321 default from the Config module. For example, to check whether the
1322 platform has the C<lstat> call, check C<$Config{d_lstat}>. See
1323 L<Config> for a full description of available variables.
1325 =head2 Alphabetical Listing of Perl Functions
1335 C<-r>, C<-w>, and C<-x> have a limited meaning only; directories
1336 and applications are executable, and there are no uid/gid
1337 considerations. C<-o> is not supported. (S<Mac OS>)
1339 C<-r>, C<-w>, C<-x>, and C<-o> tell whether the file is accessible,
1340 which may not reflect UIC-based file protections. (VMS)
1342 C<-s> returns the size of the data fork, not the total size of data fork
1343 plus resource fork. (S<Mac OS>).
1345 C<-s> by name on an open file will return the space reserved on disk,
1346 rather than the current extent. C<-s> on an open filehandle returns the
1347 current size. (S<RISC OS>)
1349 C<-R>, C<-W>, C<-X>, C<-O> are indistinguishable from C<-r>, C<-w>,
1350 C<-x>, C<-o>. (S<Mac OS>, Win32, VMS, S<RISC OS>)
1352 C<-b>, C<-c>, C<-k>, C<-g>, C<-p>, C<-u>, C<-A> are not implemented.
1355 C<-g>, C<-k>, C<-l>, C<-p>, C<-u>, C<-A> are not particularly meaningful.
1356 (Win32, VMS, S<RISC OS>)
1358 C<-d> is true if passed a device spec without an explicit directory.
1361 C<-T> and C<-B> are implemented, but might misclassify Mac text files
1362 with foreign characters; this is the case will all platforms, but may
1363 affect S<Mac OS> often. (S<Mac OS>)
1365 C<-x> (or C<-X>) determine if a file ends in one of the executable
1366 suffixes. C<-S> is meaningless. (Win32)
1368 C<-x> (or C<-X>) determine if a file has an executable file type.
1375 Not implemented. (Win32)
1377 =item binmode FILEHANDLE
1379 Meaningless. (S<Mac OS>, S<RISC OS>)
1381 Reopens file and restores pointer; if function fails, underlying
1382 filehandle may be closed, or pointer may be in a different position.
1385 The value returned by C<tell> may be affected after the call, and
1386 the filehandle may be flushed. (Win32)
1390 Only limited meaning. Disabling/enabling write permission is mapped to
1391 locking/unlocking the file. (S<Mac OS>)
1393 Only good for changing "owner" read-write access, "group", and "other"
1394 bits are meaningless. (Win32)
1396 Only good for changing "owner" and "other" read-write access. (S<RISC OS>)
1398 Access permissions are mapped onto VOS access-control list changes. (VOS)
1402 Not implemented. (S<Mac OS>, Win32, Plan9, S<RISC OS>, VOS)
1404 Does nothing, but won't fail. (Win32)
1406 =item chroot FILENAME
1410 Not implemented. (S<Mac OS>, Win32, VMS, Plan9, S<RISC OS>, VOS, VM/ESA)
1412 =item crypt PLAINTEXT,SALT
1414 May not be available if library or source was not provided when building
1417 Not implemented. (VOS)
1421 Not implemented. (VMS, Plan9, VOS)
1423 =item dbmopen HASH,DBNAME,MODE
1425 Not implemented. (VMS, Plan9, VOS)
1429 Not useful. (S<Mac OS>, S<RISC OS>)
1431 Not implemented. (Win32)
1433 Invokes VMS debugger. (VMS)
1437 Not implemented. (S<Mac OS>)
1439 Implemented via Spawn. (VM/ESA)
1441 Does not automatically flush output handles on some platforms.
1442 (SunOS, Solaris, HP-UX)
1448 Emulates UNIX exit() (which considers C<exit 1> to indicate an error) by
1449 mapping the C<1> to SS$_ABORT (C<44>). This behavior may be overridden
1450 with the pragma C<use vmsish 'exit'>. As with the CRTL's exit()
1451 function, C<exit 0> is also mapped to an exit status of SS$_NORMAL
1452 (C<1>); this mapping cannot be overridden. Any other argument to exit()
1453 is used directly as Perl's exit status. (VMS)
1455 =item fcntl FILEHANDLE,FUNCTION,SCALAR
1457 Not implemented. (Win32, VMS)
1459 =item flock FILEHANDLE,OPERATION
1461 Not implemented (S<Mac OS>, VMS, S<RISC OS>, VOS).
1463 Available only on Windows NT (not on Windows 95). (Win32)
1467 Not implemented. (S<Mac OS>, AmigaOS, S<RISC OS>, VOS, VM/ESA)
1469 Emulated using multiple interpreters. See L<perlfork>. (Win32)
1471 Does not automatically flush output handles on some platforms.
1472 (SunOS, Solaris, HP-UX)
1476 Not implemented. (S<Mac OS>, S<RISC OS>)
1480 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS)
1484 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>)
1486 =item getpriority WHICH,WHO
1488 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS, VM/ESA)
1492 Not implemented. (S<Mac OS>, Win32)
1494 Not useful. (S<RISC OS>)
1498 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>)
1500 =item getnetbyname NAME
1502 Not implemented. (S<Mac OS>, Win32, Plan9)
1506 Not implemented. (S<Mac OS>, Win32)
1508 Not useful. (S<RISC OS>)
1512 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>)
1514 =item getnetbyaddr ADDR,ADDRTYPE
1516 Not implemented. (S<Mac OS>, Win32, Plan9)
1518 =item getprotobynumber NUMBER
1520 Not implemented. (S<Mac OS>)
1522 =item getservbyport PORT,PROTO
1524 Not implemented. (S<Mac OS>)
1528 Not implemented. (S<Mac OS>, Win32, VM/ESA)
1532 Not implemented. (S<Mac OS>, Win32, VMS, VM/ESA)
1536 Not implemented. (S<Mac OS>, Win32)
1540 Not implemented. (S<Mac OS>, Win32, Plan9)
1544 Not implemented. (S<Mac OS>, Win32, Plan9)
1548 Not implemented. (Win32, Plan9)
1552 Not implemented. (S<Mac OS>, Win32, S<RISC OS>)
1556 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>)
1558 =item sethostent STAYOPEN
1560 Not implemented. (S<Mac OS>, Win32, Plan9, S<RISC OS>)
1562 =item setnetent STAYOPEN
1564 Not implemented. (S<Mac OS>, Win32, Plan9, S<RISC OS>)
1566 =item setprotoent STAYOPEN
1568 Not implemented. (S<Mac OS>, Win32, Plan9, S<RISC OS>)
1570 =item setservent STAYOPEN
1572 Not implemented. (Plan9, Win32, S<RISC OS>)
1576 Not implemented. (S<Mac OS>, MPE/iX, VM/ESA, Win32)
1580 Not implemented. (S<Mac OS>, MPE/iX, S<RISC OS>, VM/ESA, VMS, Win32)
1584 Not implemented. (S<Mac OS>, Win32)
1588 Not implemented. (S<Mac OS>, Win32, Plan9)
1592 Not implemented. (S<Mac OS>, Win32, Plan9)
1596 Not implemented. (Plan9, Win32)
1598 =item getsockopt SOCKET,LEVEL,OPTNAME
1600 Not implemented. (Plan9)
1606 This operator is implemented via the File::Glob extension on most
1607 platforms. See L<File::Glob> for portability information.
1609 =item ioctl FILEHANDLE,FUNCTION,SCALAR
1611 Not implemented. (VMS)
1613 Available only for socket handles, and it does what the ioctlsocket() call
1614 in the Winsock API does. (Win32)
1616 Available only for socket handles. (S<RISC OS>)
1618 =item kill SIGNAL, LIST
1620 C<kill(0, LIST)> is implemented for the sake of taint checking;
1621 use with other signals is unimplemented. (S<Mac OS>)
1623 Not implemented, hence not useful for taint checking. (S<RISC OS>)
1625 C<kill()> doesn't have the semantics of C<raise()>, i.e. it doesn't send
1626 a signal to the identified process like it does on Unix platforms.
1627 Instead C<kill($sig, $pid)> terminates the process identified by $pid,
1628 and makes it exit immediately with exit status $sig. As in Unix, if
1629 $sig is 0 and the specified process exists, it returns true without
1630 actually terminating it. (Win32)
1632 =item link OLDFILE,NEWFILE
1634 Not implemented. (S<Mac OS>, MPE/iX, VMS, S<RISC OS>)
1636 Link count not updated because hard links are not quite that hard
1637 (They are sort of half-way between hard and soft links). (AmigaOS)
1639 Hard links are implemented on Win32 (Windows NT and Windows 2000)
1642 =item lstat FILEHANDLE
1648 Not implemented. (VMS, S<RISC OS>)
1650 Return values (especially for device and inode) may be bogus. (Win32)
1652 =item msgctl ID,CMD,ARG
1654 =item msgget KEY,FLAGS
1656 =item msgsnd ID,MSG,FLAGS
1658 =item msgrcv ID,VAR,SIZE,TYPE,FLAGS
1660 Not implemented. (S<Mac OS>, Win32, VMS, Plan9, S<RISC OS>, VOS)
1662 =item open FILEHANDLE,EXPR
1664 =item open FILEHANDLE
1666 The C<|> variants are supported only if ToolServer is installed.
1669 open to C<|-> and C<-|> are unsupported. (S<Mac OS>, Win32, S<RISC OS>)
1671 Opening a process does not automatically flush output handles on some
1672 platforms. (SunOS, Solaris, HP-UX)
1674 =item pipe READHANDLE,WRITEHANDLE
1676 Very limited functionality. (MiNT)
1682 Not implemented. (Win32, VMS, S<RISC OS>)
1684 =item select RBITS,WBITS,EBITS,TIMEOUT
1686 Only implemented on sockets. (Win32, VMS)
1688 Only reliable on sockets. (S<RISC OS>)
1690 Note that the C<select FILEHANDLE> form is generally portable.
1692 =item semctl ID,SEMNUM,CMD,ARG
1694 =item semget KEY,NSEMS,FLAGS
1696 =item semop KEY,OPSTRING
1698 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS)
1702 Not implemented. (MPE/iX, Win32)
1704 =item setpgrp PID,PGRP
1706 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS)
1708 =item setpriority WHICH,WHO,PRIORITY
1710 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS)
1714 Not implemented. (MPE/iX, Win32)
1716 =item setsockopt SOCKET,LEVEL,OPTNAME,OPTVAL
1718 Not implemented. (Plan9)
1720 =item shmctl ID,CMD,ARG
1722 =item shmget KEY,SIZE,FLAGS
1724 =item shmread ID,VAR,POS,SIZE
1726 =item shmwrite ID,STRING,POS,SIZE
1728 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS)
1730 =item sockatmark SOCKET
1732 A relatively recent addition to socket functions, may not
1733 be implemented even in UNIX platforms.
1735 =item socketpair SOCKET1,SOCKET2,DOMAIN,TYPE,PROTOCOL
1737 Not implemented. (Win32, VMS, S<RISC OS>, VOS, VM/ESA)
1739 =item stat FILEHANDLE
1745 Platforms that do not have rdev, blksize, or blocks will return these
1746 as '', so numeric comparison or manipulation of these fields may cause
1747 'not numeric' warnings.
1749 mtime and atime are the same thing, and ctime is creation time instead of
1750 inode change time. (S<Mac OS>)
1752 device and inode are not meaningful. (Win32)
1754 device and inode are not necessarily reliable. (VMS)
1756 mtime, atime and ctime all return the last modification time. Device and
1757 inode are not necessarily reliable. (S<RISC OS>)
1759 dev, rdev, blksize, and blocks are not available. inode is not
1760 meaningful and will differ between stat calls on the same file. (os2)
1762 some versions of cygwin when doing a stat("foo") and if not finding it
1763 may then attempt to stat("foo.exe") (Cygwin)
1765 =item symlink OLDFILE,NEWFILE
1767 Not implemented. (Win32, VMS, S<RISC OS>)
1771 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS, VM/ESA)
1773 =item sysopen FILEHANDLE,FILENAME,MODE,PERMS
1775 The traditional "0", "1", and "2" MODEs are implemented with different
1776 numeric values on some systems. The flags exported by C<Fcntl>
1777 (O_RDONLY, O_WRONLY, O_RDWR) should work everywhere though. (S<Mac
1778 OS>, OS/390, VM/ESA)
1782 In general, do not assume the UNIX/POSIX semantics that you can shift
1783 the C<$?> left by eight to get the exit value, or that C<$? & 127>
1784 would give you the number of the signal that terminated the program,
1785 or that C<$? & 128> would test true if the program was terminated by a
1786 coredump. Instead, use the POSIX W*() interfaces: for example, use
1787 WIFEXITED($?) an WEXITVALUE($?) to test for a normal exit and the exit
1788 value, and WIFSIGNALED($?) and WTERMSIG($?) for a signal exit and the
1789 signal. Core dumping is not a portable concept so there's no portable
1790 way to test for that.
1792 Only implemented if ToolServer is installed. (S<Mac OS>)
1794 As an optimization, may not call the command shell specified in
1795 C<$ENV{PERL5SHELL}>. C<system(1, @args)> spawns an external
1796 process and immediately returns its process designator, without
1797 waiting for it to terminate. Return value may be used subsequently
1798 in C<wait> or C<waitpid>. Failure to spawn() a subprocess is indicated
1799 by setting $? to "255 << 8". C<$?> is set in a way compatible with
1800 Unix (i.e. the exitstatus of the subprocess is obtained by "$? >> 8",
1801 as described in the documentation). (Win32)
1803 There is no shell to process metacharacters, and the native standard is
1804 to pass a command line terminated by "\n" "\r" or "\0" to the spawned
1805 program. Redirection such as C<< > foo >> is performed (if at all) by
1806 the run time library of the spawned program. C<system> I<list> will call
1807 the Unix emulation library's C<exec> emulation, which attempts to provide
1808 emulation of the stdin, stdout, stderr in force in the parent, providing
1809 the child program uses a compatible version of the emulation library.
1810 I<scalar> will call the native command line direct and no such emulation
1811 of a child Unix program will exists. Mileage B<will> vary. (S<RISC OS>)
1813 Far from being POSIX compliant. Because there may be no underlying
1814 /bin/sh tries to work around the problem by forking and execing the
1815 first token in its argument string. Handles basic redirection
1816 ("<" or ">") on its own behalf. (MiNT)
1818 Does not automatically flush output handles on some platforms.
1819 (SunOS, Solaris, HP-UX)
1821 The return value is POSIX-like (shifted up by 8 bits), which only allows
1822 room for a made-up value derived from the severity bits of the native
1823 32-bit condition code (unless overridden by C<use vmsish 'status'>).
1824 For more details see L<perlvms/$?>. (VMS)
1828 Only the first entry returned is nonzero. (S<Mac OS>)
1830 "cumulative" times will be bogus. On anything other than Windows NT
1831 or Windows 2000, "system" time will be bogus, and "user" time is
1832 actually the time returned by the clock() function in the C runtime
1835 Not useful. (S<RISC OS>)
1837 =item truncate FILEHANDLE,LENGTH
1839 =item truncate EXPR,LENGTH
1841 Not implemented. (Older versions of VMS)
1843 Truncation to zero-length only. (VOS)
1845 If a FILEHANDLE is supplied, it must be writable and opened in append
1846 mode (i.e., use C<<< open(FH, '>>filename') >>>
1847 or C<sysopen(FH,...,O_APPEND|O_RDWR)>. If a filename is supplied, it
1848 should not be held open elsewhere. (Win32)
1854 Returns undef where unavailable, as of version 5.005.
1856 C<umask> works but the correct permissions are set only when the file
1857 is finally closed. (AmigaOS)
1861 Only the modification time is updated. (S<BeOS>, S<Mac OS>, VMS, S<RISC OS>)
1863 May not behave as expected. Behavior depends on the C runtime
1864 library's implementation of utime(), and the filesystem being
1865 used. The FAT filesystem typically does not support an "access
1866 time" field, and it may limit timestamps to a granularity of
1867 two seconds. (Win32)
1871 =item waitpid PID,FLAGS
1873 Not implemented. (S<Mac OS>, VOS)
1875 Can only be applied to process handles returned for processes spawned
1876 using C<system(1, ...)> or pseudo processes created with C<fork()>. (Win32)
1878 Not useful. (S<RISC OS>)
1886 =item v1.48, 02 February 2001
1888 Various updates from perl5-porters over the past year, supported
1889 platforms update from Jarkko Hietaniemi.
1891 =item v1.47, 22 March 2000
1893 Various cleanups from Tom Christiansen, including migration of
1894 long platform listings from L<perl>.
1896 =item v1.46, 12 February 2000
1898 Updates for VOS and MPE/iX. (Peter Prymmer) Other small changes.
1900 =item v1.45, 20 December 1999
1902 Small changes from 5.005_63 distribution, more changes to EBCDIC info.
1904 =item v1.44, 19 July 1999
1906 A bunch of updates from Peter Prymmer for C<$^O> values,
1907 endianness, File::Spec, VMS, BS2000, OS/400.
1909 =item v1.43, 24 May 1999
1911 Added a lot of cleaning up from Tom Christiansen.
1913 =item v1.42, 22 May 1999
1915 Added notes about tests, sprintf/printf, and epoch offsets.
1917 =item v1.41, 19 May 1999
1919 Lots more little changes to formatting and content.
1921 Added a bunch of C<$^O> and related values
1922 for various platforms; fixed mail and web addresses, and added
1923 and changed miscellaneous notes. (Peter Prymmer)
1925 =item v1.40, 11 April 1999
1927 Miscellaneous changes.
1929 =item v1.39, 11 February 1999
1931 Changes from Jarkko and EMX URL fixes Michael Schwern. Additional
1932 note about newlines added.
1934 =item v1.38, 31 December 1998
1936 More changes from Jarkko.
1938 =item v1.37, 19 December 1998
1940 More minor changes. Merge two separate version 1.35 documents.
1942 =item v1.36, 9 September 1998
1944 Updated for Stratus VOS. Also known as version 1.35.
1946 =item v1.35, 13 August 1998
1948 Integrate more minor changes, plus addition of new sections under
1949 L<"ISSUES">: L<"Numbers endianness and Width">,
1950 L<"Character sets and character encoding">,
1951 L<"Internationalisation">.
1953 =item v1.33, 06 August 1998
1955 Integrate more minor changes.
1957 =item v1.32, 05 August 1998
1959 Integrate more minor changes.
1961 =item v1.30, 03 August 1998
1963 Major update for RISC OS, other minor changes.
1965 =item v1.23, 10 July 1998
1967 First public release with perl5.005.
1971 =head1 Supported Platforms
1973 As of early 2001 (the Perl releases 5.6.1 and 5.7.1), the following
1974 platforms are able to build Perl from the standard source code
1975 distribution available at http://www.cpan.org/src/index.html
1998 Tru64 UNIX (DEC OSF/1, Digital UNIX)
2004 1) in DOS mode either the DOS or OS/2 ports can be used
2005 2) Mac OS Classic (pre-X) is almost 5.6.1-ready; building from
2006 the source does work with 5.6.1, but additional MacOS specific
2007 source code is needed for a complete build. See the web
2008 site http://dev.macperl.org/ for more information.
2009 3) compilers: Borland, Cygwin, Mingw32 EGCS/GCC, VC++
2011 The following platforms worked for the previous releases (5.6.0 and 5.7.0),
2012 but we did not manage to test these in time for the 5.7.1 release.
2013 There is a very good chance that these will work fine with the 5.7.1.
2031 The following platform worked for the 5.005_03 major release but not
2032 for 5.6.0. Standardization on UTF-8 as the internal string
2033 representation in 5.6.0 and 5.6.1 introduced incompatibilities in this
2034 EBCDIC platform. While Perl 5.7.1 will build on this platform some
2035 regression tests may fail and the C<use utf8;> pragma typically
2036 introduces text handling errors.
2040 1) previously known as MVS, about to become z/OS.
2042 Strongly related to the OS/390 platform by also being EBCDIC-based
2043 mainframe platforms are the following platforms:
2048 These are also expected to work, albeit with no UTF-8 support, under 5.6.1
2049 for the same reasons as OS/390. Contact the mailing list perl-mvs@perl.org
2052 The following platforms have been known to build Perl from source in
2053 the past (5.005_03 and earlier), but we haven't been able to verify
2054 their status for the current release, either because the
2055 hardware/software platforms are rare or because we don't have an
2056 active champion on these platforms--or both. They used to work,
2057 though, so go ahead and try compiling them, and let perlbug@perl.org
2096 Support for the following platform is planned for a future Perl release:
2100 The following platforms have their own source code distributions and
2101 binaries available via http://www.cpan.org/ports/index.html:
2107 Tandem Guardian 5.004
2109 The following platforms have only binaries available via
2110 http://www.cpan.org/ports/index.html :
2114 Acorn RISCOS 5.005_02
2118 Although we do suggest that you always build your own Perl from
2119 the source code, both for maximal configurability and for security,
2120 in case you are in a hurry you can check
2121 http://www.cpan.org/ports/index.html for binary distributions.
2125 L<perlaix>, L<perlapollo>, L<perlamiga>, L<perlbeos>, L<perlbs200>,
2126 L<perlce>, L<perlcygwin>, L<perldgux>, L<perldos>, L<perlepoc>, L<perlebcdic>,
2127 L<perlhurd>, L<perlhpux>, L<perlmachten>, L<perlmacos>, L<perlmint>,
2128 L<perlmpeix>, L<perlnetware>, L<perlos2>, L<perlos390>, L<perlplan9>,
2129 L<perlqnx>, L<perlsolaris>, L<perltru64>, L<perlunicode>,
2130 L<perlvmesa>, L<perlvms>, L<perlvos>, L<perlwin32>, and L<Win32>.
2132 =head1 AUTHORS / CONTRIBUTORS
2134 Abigail <abigail@foad.org>,
2135 Charles Bailey <bailey@newman.upenn.edu>,
2136 Graham Barr <gbarr@pobox.com>,
2137 Tom Christiansen <tchrist@perl.com>,
2138 Nicholas Clark <nick@ccl4.org>,
2139 Thomas Dorner <Thomas.Dorner@start.de>,
2140 Andy Dougherty <doughera@lafayette.edu>,
2141 Dominic Dunlop <domo@computer.org>,
2142 Neale Ferguson <neale@vma.tabnsw.com.au>,
2143 David J. Fiander <davidf@mks.com>,
2144 Paul Green <Paul_Green@stratus.com>,
2145 M.J.T. Guy <mjtg@cam.ac.uk>,
2146 Jarkko Hietaniemi <jhi@iki.fi>,
2147 Luther Huffman <lutherh@stratcom.com>,
2148 Nick Ing-Simmons <nick@ing-simmons.net>,
2149 Andreas J. KE<ouml>nig <a.koenig@mind.de>,
2150 Markus Laker <mlaker@contax.co.uk>,
2151 Andrew M. Langmead <aml@world.std.com>,
2152 Larry Moore <ljmoore@freespace.net>,
2153 Paul Moore <Paul.Moore@uk.origin-it.com>,
2154 Chris Nandor <pudge@pobox.com>,
2155 Matthias Neeracher <neeri@iis.ee.ethz.ch>,
2156 Philip Newton <pne@cpan.org>,
2157 Gary Ng <71564.1743@CompuServe.COM>,
2158 Tom Phoenix <rootbeer@teleport.com>,
2159 AndrE<eacute> Pirard <A.Pirard@ulg.ac.be>,
2160 Peter Prymmer <pvhp@forte.com>,
2161 Hugo van der Sanden <hv@crypt0.demon.co.uk>,
2162 Gurusamy Sarathy <gsar@activestate.com>,
2163 Paul J. Schinder <schinder@pobox.com>,
2164 Michael G Schwern <schwern@pobox.com>,
2165 Dan Sugalski <dan@sidhe.org>,
2166 Nathan Torkington <gnat@frii.com>.
2170 Version 1.50, last modified 10 Jul 2001