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 Don't count on signals or C<%SIG> for anything.
438 Don't count on filename globbing. Use C<opendir>, C<readdir>, and
441 Don't count on per-program environment variables, or per-program current
444 Don't count on specific values of C<$!>, neither numeric nor
445 especially the strings values-- users may switch their locales causing
446 error messages to be translated into their languages. If you can
447 trust a POSIXish environment, you can portably use the symbols defined
448 by the Errno module, like ENOENT. And don't trust on the values of C<$!>
449 at all except immediately after a failed system call.
451 =head2 Command names versus file pathnames
453 Don't assume that the name used to invoke a command or program with
454 C<system> or C<exec> can also be used to test for the existence of the
455 file that holds the executable code for that command or program.
456 First, many systems have "internal" commands that are built-in to the
457 shell or OS and while these commands can be invoked, there is no
458 corresponding file. Second, some operating systems (e.g., Cygwin,
459 DJGPP, OS/2, and VOS) have required suffixes for executable files;
460 these suffixes are generally permitted on the command name but are not
461 required. Thus, a command like "perl" might exist in a file named
462 "perl", "perl.exe", or "perl.pm", depending on the operating system.
463 The variable "_exe" in the Config module holds the executable suffix,
464 if any. Third, the VMS port carefully sets up $^X and
465 $Config{perlpath} so that no further processing is required. This is
466 just as well, because the matching regular expression used below would
467 then have to deal with a possible trailing version number in the VMS
470 To convert $^X to a file pathname, taking account of the requirements
471 of the various operating system possibilities, say:
476 {$thisperl .= $Config{_exe} unless $thisperl =~ m/$Config{_exe}$/i;}
478 To convert $Config{perlpath} to a file pathname, say:
481 $thisperl = $Config{perlpath};
483 {$thisperl .= $Config{_exe} unless $thisperl =~ m/$Config{_exe}$/i;}
487 Don't assume that you can reach the public Internet.
489 Don't assume that there is only one way to get through firewalls
490 to the public Internet.
492 Don't assume that you can reach outside world through any other port
493 than 80, or some web proxy. ftp is blocked by many firewalls.
495 Don't assume that you can send email by connecting to the local SMTP port.
497 Don't assume that you can reach yourself or any node by the name
498 'localhost'. The same goes for '127.0.0.1'. You will have to try both.
500 Don't assume that the host has only one network card, or that it
501 can't bind to many virtual IP addresses.
503 Don't assume a particular network device name.
505 Don't assume a particular set of ioctl()s will work.
507 Don't assume that you can ping hosts and get replies.
509 Don't assume that any particular port (service) will respond.
511 Don't assume that Sys::Hostname (or any other API or command)
512 returns either a fully qualified hostname or a non-qualified hostname:
513 it all depends on how the system had been configured. Also remember
514 things like DHCP and NAT-- the hostname you get back might not be very
517 All the above "don't":s may look daunting, and they are -- but the key
518 is to degrade gracefully if one cannot reach the particular network
519 service one wants. Croaking or hanging do not look very professional.
521 =head2 Interprocess Communication (IPC)
523 In general, don't directly access the system in code meant to be
524 portable. That means, no C<system>, C<exec>, C<fork>, C<pipe>,
525 C<``>, C<qx//>, C<open> with a C<|>, nor any of the other things
526 that makes being a perl hacker worth being.
528 Commands that launch external processes are generally supported on
529 most platforms (though many of them do not support any type of
530 forking). The problem with using them arises from what you invoke
531 them on. External tools are often named differently on different
532 platforms, may not be available in the same location, might accept
533 different arguments, can behave differently, and often present their
534 results in a platform-dependent way. Thus, you should seldom depend
535 on them to produce consistent results. (Then again, if you're calling
536 I<netstat -a>, you probably don't expect it to run on both Unix and CP/M.)
538 One especially common bit of Perl code is opening a pipe to B<sendmail>:
540 open(MAIL, '|/usr/lib/sendmail -t')
541 or die "cannot fork sendmail: $!";
543 This is fine for systems programming when sendmail is known to be
544 available. But it is not fine for many non-Unix systems, and even
545 some Unix systems that may not have sendmail installed. If a portable
546 solution is needed, see the various distributions on CPAN that deal
547 with it. Mail::Mailer and Mail::Send in the MailTools distribution are
548 commonly used, and provide several mailing methods, including mail,
549 sendmail, and direct SMTP (via Net::SMTP) if a mail transfer agent is
550 not available. Mail::Sendmail is a standalone module that provides
551 simple, platform-independent mailing.
553 The Unix System V IPC (C<msg*(), sem*(), shm*()>) is not available
554 even on all Unix platforms.
556 Do not use either the bare result of C<pack("N", 10, 20, 30, 40)> or
557 bare v-strings (such as C<v10.20.30.40>) to represent IPv4 addresses:
558 both forms just pack the four bytes into network order. That this
559 would be equal to the C language C<in_addr> struct (which is what the
560 socket code internally uses) is not guaranteed. To be portable use
561 the routines of the Socket extension, such as C<inet_aton()>,
562 C<inet_ntoa()>, and C<sockaddr_in()>.
564 The rule of thumb for portable code is: Do it all in portable Perl, or
565 use a module (that may internally implement it with platform-specific
566 code, but expose a common interface).
568 =head2 External Subroutines (XS)
570 XS code can usually be made to work with any platform, but dependent
571 libraries, header files, etc., might not be readily available or
572 portable, or the XS code itself might be platform-specific, just as Perl
573 code might be. If the libraries and headers are portable, then it is
574 normally reasonable to make sure the XS code is portable, too.
576 A different type of portability issue arises when writing XS code:
577 availability of a C compiler on the end-user's system. C brings
578 with it its own portability issues, and writing XS code will expose
579 you to some of those. Writing purely in Perl is an easier way to
582 =head2 Standard Modules
584 In general, the standard modules work across platforms. Notable
585 exceptions are the CPAN module (which currently makes connections to external
586 programs that may not be available), platform-specific modules (like
587 ExtUtils::MM_VMS), and DBM modules.
589 There is no one DBM module available on all platforms.
590 SDBM_File and the others are generally available on all Unix and DOSish
591 ports, but not in MacPerl, where only NBDM_File and DB_File are
594 The good news is that at least some DBM module should be available, and
595 AnyDBM_File will use whichever module it can find. Of course, then
596 the code needs to be fairly strict, dropping to the greatest common
597 factor (e.g., not exceeding 1K for each record), so that it will
598 work with any DBM module. See L<AnyDBM_File> for more details.
602 The system's notion of time of day and calendar date is controlled in
603 widely different ways. Don't assume the timezone is stored in C<$ENV{TZ}>,
604 and even if it is, don't assume that you can control the timezone through
605 that variable. Don't assume anything about the three-letter timezone
606 abbreviations (for example that MST would be the Mountain Standard Time,
607 it's been known to stand for Moscow Standard Time). If you need to
608 use timezones, express them in some unambiguous format like the
609 exact number of minutes offset from UTC, or the POSIX timezone
612 Don't assume that the epoch starts at 00:00:00, January 1, 1970,
613 because that is OS- and implementation-specific. It is better to
614 store a date in an unambiguous representation. The ISO 8601 standard
615 defines YYYY-MM-DD as the date format, or YYYY-MM-DDTHH-MM-SS
616 (that's a literal "T" separating the date from the time).
617 Please do use the ISO 8601 instead of making us to guess what
618 date 02/03/04 might be. ISO 8601 even sorts nicely as-is.
619 A text representation (like "1987-12-18") can be easily converted
620 into an OS-specific value using a module like Date::Parse.
621 An array of values, such as those returned by C<localtime>, can be
622 converted to an OS-specific representation using Time::Local.
624 When calculating specific times, such as for tests in time or date modules,
625 it may be appropriate to calculate an offset for the epoch.
628 $offset = Time::Local::timegm(0, 0, 0, 1, 0, 70);
630 The value for C<$offset> in Unix will be C<0>, but in Mac OS will be
631 some large number. C<$offset> can then be added to a Unix time value
632 to get what should be the proper value on any system.
634 On Windows (at least), you shouldn't pass a negative value to C<gmtime> or
637 =head2 Character sets and character encoding
639 Assume very little about character sets.
641 Assume nothing about numerical values (C<ord>, C<chr>) of characters.
642 Do not use explicit code point ranges (like \xHH-\xHH); use for
643 example symbolic character classes like C<[:print:]>.
645 Do not assume that the alphabetic characters are encoded contiguously
646 (in the numeric sense). There may be gaps.
648 Do not assume anything about the ordering of the characters.
649 The lowercase letters may come before or after the uppercase letters;
650 the lowercase and uppercase may be interlaced so that both "a" and "A"
651 come before "b"; the accented and other international characters may
652 be interlaced so that E<auml> comes before "b".
654 =head2 Internationalisation
656 If you may assume POSIX (a rather large assumption), you may read
657 more about the POSIX locale system from L<perllocale>. The locale
658 system at least attempts to make things a little bit more portable,
659 or at least more convenient and native-friendly for non-English
660 users. The system affects character sets and encoding, and date
661 and time formatting--amongst other things.
663 If you really want to be international, you should consider Unicode.
664 See L<perluniintro> and L<perlunicode> for more information.
666 If you want to use non-ASCII bytes (outside the bytes 0x00..0x7f) in
667 the "source code" of your code, to be portable you have to be explicit
668 about what bytes they are. Someone might for example be using your
669 code under a UTF-8 locale, in which case random native bytes might be
670 illegal ("Malformed UTF-8 ...") This means that for example embedding
671 ISO 8859-1 bytes beyond 0x7f into your strings might cause trouble
672 later. If the bytes are native 8-bit bytes, you can use the C<bytes>
673 pragma. If the bytes are in a string (regular expression being a
674 curious string), you can often also use the C<\xHH> notation instead
675 of embedding the bytes as-is. If they are in some particular legacy
676 encoding (ether single-byte or something more complicated), you can
677 use the C<encoding> pragma. (If you want to write your code in UTF-8,
678 you can use either the C<utf8> pragma, or the C<encoding> pragma.)
679 The C<bytes> and C<utf8> pragmata are available since Perl 5.6.0, and
680 the C<encoding> pragma since Perl 5.8.0.
682 =head2 System Resources
684 If your code is destined for systems with severely constrained (or
685 missing!) virtual memory systems then you want to be I<especially> mindful
686 of avoiding wasteful constructs such as:
688 # NOTE: this is no longer "bad" in perl5.005
689 for (0..10000000) {} # bad
690 for (my $x = 0; $x <= 10000000; ++$x) {} # good
692 @lines = <VERY_LARGE_FILE>; # bad
694 while (<FILE>) {$file .= $_} # sometimes bad
695 $file = join('', <FILE>); # better
697 The last two constructs may appear unintuitive to most people. The
698 first repeatedly grows a string, whereas the second allocates a
699 large chunk of memory in one go. On some systems, the second is
700 more efficient that the first.
704 Most multi-user platforms provide basic levels of security, usually
705 implemented at the filesystem level. Some, however, do
706 not-- unfortunately. Thus the notion of user id, or "home" directory,
707 or even the state of being logged-in, may be unrecognizable on many
708 platforms. If you write programs that are security-conscious, it
709 is usually best to know what type of system you will be running
710 under so that you can write code explicitly for that platform (or
713 Don't assume the UNIX filesystem access semantics: the operating
714 system or the filesystem may be using some ACL systems, which are
715 richer languages than the usual rwx. Even if the rwx exist,
716 their semantics might be different.
718 (From security viewpoint testing for permissions before attempting to
719 do something is silly anyway: if one tries this, there is potential
720 for race conditions-- someone or something might change the
721 permissions between the permissions check and the actual operation.
722 Just try the operation.)
724 Don't assume the UNIX user and group semantics: especially, don't
725 expect the C<< $< >> and C<< $> >> (or the C<$(> and C<$)>) to work
726 for switching identities (or memberships).
728 Don't assume set-uid and set-gid semantics. (And even if you do,
729 think twice: set-uid and set-gid are a known can of security worms.)
733 For those times when it is necessary to have platform-specific code,
734 consider keeping the platform-specific code in one place, making porting
735 to other platforms easier. Use the Config module and the special
736 variable C<$^O> to differentiate platforms, as described in
739 Be careful in the tests you supply with your module or programs.
740 Module code may be fully portable, but its tests might not be. This
741 often happens when tests spawn off other processes or call external
742 programs to aid in the testing, or when (as noted above) the tests
743 assume certain things about the filesystem and paths. Be careful not
744 to depend on a specific output style for errors, such as when checking
745 C<$!> after a failed system call. Using C<$!> for anything else than
746 displaying it as output is doubtful (though see the Errno module for
747 testing reasonably portably for error value). Some platforms expect
748 a certain output format, and Perl on those platforms may have been
749 adjusted accordingly. Most specifically, don't anchor a regex when
750 testing an error value.
754 Modules uploaded to CPAN are tested by a variety of volunteers on
755 different platforms. These CPAN testers are notified by mail of each
756 new upload, and reply to the list with PASS, FAIL, NA (not applicable to
757 this platform), or UNKNOWN (unknown), along with any relevant notations.
759 The purpose of the testing is twofold: one, to help developers fix any
760 problems in their code that crop up because of lack of testing on other
761 platforms; two, to provide users with information about whether
762 a given module works on a given platform.
770 Mailing list: cpan-testers@perl.org
774 Testing results: http://testers.cpan.org/
780 As of version 5.002, Perl is built with a C<$^O> variable that
781 indicates the operating system it was built on. This was implemented
782 to help speed up code that would otherwise have to C<use Config>
783 and use the value of C<$Config{osname}>. Of course, to get more
784 detailed information about the system, looking into C<%Config> is
785 certainly recommended.
787 C<%Config> cannot always be trusted, however, because it was built
788 at compile time. If perl was built in one place, then transferred
789 elsewhere, some values may be wrong. The values may even have been
790 edited after the fact.
794 Perl works on a bewildering variety of Unix and Unix-like platforms (see
795 e.g. most of the files in the F<hints/> directory in the source code kit).
796 On most of these systems, the value of C<$^O> (hence C<$Config{'osname'}>,
797 too) is determined either by lowercasing and stripping punctuation from the
798 first field of the string returned by typing C<uname -a> (or a similar command)
799 at the shell prompt or by testing the file system for the presence of
800 uniquely named files such as a kernel or header file. Here, for example,
801 are a few of the more popular Unix flavors:
803 uname $^O $Config{'archname'}
804 --------------------------------------------
806 BSD/OS bsdos i386-bsdos
808 dgux dgux AViiON-dgux
809 DYNIX/ptx dynixptx i386-dynixptx
810 FreeBSD freebsd freebsd-i386
811 Linux linux arm-linux
812 Linux linux i386-linux
813 Linux linux i586-linux
814 Linux linux ppc-linux
815 HP-UX hpux PA-RISC1.1
817 Mac OS X darwin darwin
818 MachTen PPC machten powerpc-machten
820 NeXT 4 next OPENSTEP-Mach
821 openbsd openbsd i386-openbsd
822 OSF1 dec_osf alpha-dec_osf
823 reliantunix-n svr4 RM400-svr4
824 SCO_SV sco_sv i386-sco_sv
825 SINIX-N svr4 RM400-svr4
826 sn4609 unicos CRAY_C90-unicos
827 sn6521 unicosmk t3e-unicosmk
828 sn9617 unicos CRAY_J90-unicos
829 SunOS solaris sun4-solaris
830 SunOS solaris i86pc-solaris
831 SunOS4 sunos sun4-sunos
833 Because the value of C<$Config{archname}> may depend on the
834 hardware architecture, it can vary more than the value of C<$^O>.
836 =head2 DOS and Derivatives
838 Perl has long been ported to Intel-style microcomputers running under
839 systems like PC-DOS, MS-DOS, OS/2, and most Windows platforms you can
840 bring yourself to mention (except for Windows CE, if you count that).
841 Users familiar with I<COMMAND.COM> or I<CMD.EXE> style shells should
842 be aware that each of these file specifications may have subtle
845 $filespec0 = "c:/foo/bar/file.txt";
846 $filespec1 = "c:\\foo\\bar\\file.txt";
847 $filespec2 = 'c:\foo\bar\file.txt';
848 $filespec3 = 'c:\\foo\\bar\\file.txt';
850 System calls accept either C</> or C<\> as the path separator.
851 However, many command-line utilities of DOS vintage treat C</> as
852 the option prefix, so may get confused by filenames containing C</>.
853 Aside from calling any external programs, C</> will work just fine,
854 and probably better, as it is more consistent with popular usage,
855 and avoids the problem of remembering what to backwhack and what
858 The DOS FAT filesystem can accommodate only "8.3" style filenames. Under
859 the "case-insensitive, but case-preserving" HPFS (OS/2) and NTFS (NT)
860 filesystems you may have to be careful about case returned with functions
861 like C<readdir> or used with functions like C<open> or C<opendir>.
863 DOS also treats several filenames as special, such as AUX, PRN,
864 NUL, CON, COM1, LPT1, LPT2, etc. Unfortunately, sometimes these
865 filenames won't even work if you include an explicit directory
866 prefix. It is best to avoid such filenames, if you want your code
867 to be portable to DOS and its derivatives. It's hard to know what
868 these all are, unfortunately.
870 Users of these operating systems may also wish to make use of
871 scripts such as I<pl2bat.bat> or I<pl2cmd> to
872 put wrappers around your scripts.
874 Newline (C<\n>) is translated as C<\015\012> by STDIO when reading from
875 and writing to files (see L<"Newlines">). C<binmode(FILEHANDLE)>
876 will keep C<\n> translated as C<\012> for that filehandle. Since it is a
877 no-op on other systems, C<binmode> should be used for cross-platform code
878 that deals with binary data. That's assuming you realize in advance
879 that your data is in binary. General-purpose programs should
880 often assume nothing about their data.
882 The C<$^O> variable and the C<$Config{archname}> values for various
883 DOSish perls are as follows:
885 OS $^O $Config{archname} ID Version
886 --------------------------------------------------------
890 Windows 3.1 ? ? 0 3 01
891 Windows 95 MSWin32 MSWin32-x86 1 4 00
892 Windows 98 MSWin32 MSWin32-x86 1 4 10
893 Windows ME MSWin32 MSWin32-x86 1 ?
894 Windows NT MSWin32 MSWin32-x86 2 4 xx
895 Windows NT MSWin32 MSWin32-ALPHA 2 4 xx
896 Windows NT MSWin32 MSWin32-ppc 2 4 xx
897 Windows 2000 MSWin32 MSWin32-x86 2 5 00
898 Windows XP MSWin32 MSWin32-x86 2 5 01
899 Windows 2003 MSWin32 MSWin32-x86 2 5 02
900 Windows CE MSWin32 ? 3
903 The various MSWin32 Perl's can distinguish the OS they are running on
904 via the value of the fifth element of the list returned from
905 Win32::GetOSVersion(). For example:
907 if ($^O eq 'MSWin32') {
908 my @os_version_info = Win32::GetOSVersion();
909 print +('3.1','95','NT')[$os_version_info[4]],"\n";
912 There are also Win32::IsWinNT() and Win32::IsWin95(), try C<perldoc Win32>,
913 and as of libwin32 0.19 (not part of the core Perl distribution)
914 Win32::GetOSName(). The very portable POSIX::uname() will work too:
916 c:\> perl -MPOSIX -we "print join '|', uname"
917 Windows NT|moonru|5.0|Build 2195 (Service Pack 2)|x86
925 The djgpp environment for DOS, http://www.delorie.com/djgpp/
930 The EMX environment for DOS, OS/2, etc. emx@iaehv.nl,
931 http://www.leo.org/pub/comp/os/os2/leo/gnu/emx+gcc/index.html or
932 ftp://hobbes.nmsu.edu/pub/os2/dev/emx/ Also L<perlos2>.
936 Build instructions for Win32 in L<perlwin32>, or under the Cygnus environment
941 The C<Win32::*> modules in L<Win32>.
945 The ActiveState Pages, http://www.activestate.com/
949 The Cygwin environment for Win32; F<README.cygwin> (installed
950 as L<perlcygwin>), http://www.cygwin.com/
954 The U/WIN environment for Win32,
955 http://www.research.att.com/sw/tools/uwin/
959 Build instructions for OS/2, L<perlos2>
965 Any module requiring XS compilation is right out for most people, because
966 MacPerl is built using non-free (and non-cheap!) compilers. Some XS
967 modules that can work with MacPerl are built and distributed in binary
970 Directories are specified as:
972 volume:folder:file for absolute pathnames
973 volume:folder: for absolute pathnames
974 :folder:file for relative pathnames
975 :folder: for relative pathnames
976 :file for relative pathnames
977 file for relative pathnames
979 Files are stored in the directory in alphabetical order. Filenames are
980 limited to 31 characters, and may include any character except for
981 null and C<:>, which is reserved as the path separator.
983 Instead of C<flock>, see C<FSpSetFLock> and C<FSpRstFLock> in the
984 Mac::Files module, or C<chmod(0444, ...)> and C<chmod(0666, ...)>.
986 In the MacPerl application, you can't run a program from the command line;
987 programs that expect C<@ARGV> to be populated can be edited with something
988 like the following, which brings up a dialog box asking for the command
992 @ARGV = split /\s+/, MacPerl::Ask('Arguments?');
995 A MacPerl script saved as a "droplet" will populate C<@ARGV> with the full
996 pathnames of the files dropped onto the script.
998 Mac users can run programs under a type of command line interface
999 under MPW (Macintosh Programmer's Workshop, a free development
1000 environment from Apple). MacPerl was first introduced as an MPW
1001 tool, and MPW can be used like a shell:
1003 perl myscript.plx some arguments
1005 ToolServer is another app from Apple that provides access to MPW tools
1006 from MPW and the MacPerl app, which allows MacPerl programs to use
1007 C<system>, backticks, and piped C<open>.
1009 "S<Mac OS>" is the proper name for the operating system, but the value
1010 in C<$^O> is "MacOS". To determine architecture, version, or whether
1011 the application or MPW tool version is running, check:
1013 $is_app = $MacPerl::Version =~ /App/;
1014 $is_tool = $MacPerl::Version =~ /MPW/;
1015 ($version) = $MacPerl::Version =~ /^(\S+)/;
1016 $is_ppc = $MacPerl::Architecture eq 'MacPPC';
1017 $is_68k = $MacPerl::Architecture eq 'Mac68K';
1019 S<Mac OS X>, based on NeXT's OpenStep OS, runs MacPerl natively, under the
1020 "Classic" environment. There is no "Carbon" version of MacPerl to run
1021 under the primary Mac OS X environment. S<Mac OS X> and its Open Source
1022 version, Darwin, both run Unix perl natively.
1030 MacPerl Development, http://dev.macperl.org/ .
1034 The MacPerl Pages, http://www.macperl.com/ .
1038 The MacPerl mailing lists, http://lists.perl.org/ .
1042 MPW, ftp://ftp.apple.com/developer/Tool_Chest/Core_Mac_OS_Tools/
1048 Perl on VMS is discussed in L<perlvms> in the perl distribution.
1050 The official name of VMS as of this writing is OpenVMS.
1052 Perl on VMS can accept either VMS- or Unix-style file
1053 specifications as in either of the following:
1055 $ perl -ne "print if /perl_setup/i" SYS$LOGIN:LOGIN.COM
1056 $ perl -ne "print if /perl_setup/i" /sys$login/login.com
1058 but not a mixture of both as in:
1060 $ perl -ne "print if /perl_setup/i" sys$login:/login.com
1061 Can't open sys$login:/login.com: file specification syntax error
1063 Interacting with Perl from the Digital Command Language (DCL) shell
1064 often requires a different set of quotation marks than Unix shells do.
1067 $ perl -e "print ""Hello, world.\n"""
1070 There are several ways to wrap your perl scripts in DCL F<.COM> files, if
1071 you are so inclined. For example:
1073 $ write sys$output "Hello from DCL!"
1075 $ then perl -x 'f$environment("PROCEDURE")
1076 $ else perl -x - 'p1 'p2 'p3 'p4 'p5 'p6 'p7 'p8
1077 $ deck/dollars="__END__"
1080 print "Hello from Perl!\n";
1085 Do take care with C<$ ASSIGN/nolog/user SYS$COMMAND: SYS$INPUT> if your
1086 perl-in-DCL script expects to do things like C<< $read = <STDIN>; >>.
1088 The VMS operating system has two filesystems, known as ODS-2 and ODS-5.
1090 For ODS-2, filenames are in the format "name.extension;version". The
1091 maximum length for filenames is 39 characters, and the maximum length for
1092 extensions is also 39 characters. Version is a number from 1 to
1093 32767. Valid characters are C</[A-Z0-9$_-]/>.
1095 The ODS-2 filesystem is case-insensitive and does not preserve case.
1096 Perl simulates this by converting all filenames to lowercase internally.
1098 For ODS-5, filenames may have almost any character in them and can include
1099 Unicode characters. Characters that could be misinterpreted by the DCL
1100 shell or file parsing utilities need to be prefixed with the C<^>
1101 character, or replaced with hexadecimal characters prefixed with the
1102 C<^> character. Such prefixing is only needed with the pathnames are
1103 in VMS format in applications. Programs that can accept the UNIX format
1104 of pathnames do not need the escape characters. The maximum length for
1105 filenames is 255 characters. The ODS-5 file system can handle both
1106 a case preserved and a case sensitive mode.
1108 ODS-5 is only available on the OpenVMS for 64 bit platforms.
1110 Support for the extended file specifications is being done as optional
1111 settings to preserve backward compatibility with Perl scripts that
1112 assume the previous VMS limitations.
1114 In general routines on VMS that get a UNIX format file specification
1115 should return it in a UNIX format, and when they get a VMS format
1116 specification they should return a VMS format unless they are documented
1119 For routines that generate return a file specification, VMS allows setting
1120 if the C library which Perl is built on if it will be returned in VMS
1121 format or in UNIX format.
1123 With the ODS-2 file system, there is not much difference in syntax of
1124 filenames without paths for VMS or UNIX. With the extended character
1125 set available with ODS-5 there can be a significant difference.
1127 Because of this, existing Perl scripts written for VMS were sometimes
1128 treating VMS and UNIX filenames interchangeably. Without the extended
1129 character set enabled, this behavior will mostly be maintained for
1130 backwards compatibility.
1132 When extended characters are enabled with ODS-5, the handling of
1133 UNIX formatted file specifications is to that of a UNIX system.
1135 VMS file specifications without extensions have a trailing dot. An
1136 equivalent UNIX file specification should not show the trailing dot.
1138 The result of all of this, is that for VMS, for portable scripts, you
1139 can not depend on Perl to present the filenames in lowercase, to be
1140 case sensitive, and that the filenames could be returned in either
1143 And if a routine returns a file specification, unless it is intended to
1144 convert it, it should return it in the same format as it found it.
1146 C<readdir> by default has traditionally returned lowercased filenames.
1147 When the ODS-5 support is enabled, it will return the exact case of the
1148 filename on the disk.
1150 Files without extensions have a trailing period on them, so doing a
1151 C<readdir> in the default mode with a file named F<A.;5> will
1152 return F<a.> when VMS is (though that file could be opened with
1155 With support for extended file specifications and if C<opendir> was
1156 given a UNIX format directory, a file named F<A.;5> will return F<a>
1157 and optionally in the exact case on the disk. When C<opendir> is given
1158 a VMS format directory, then C<readdir> should return F<a.>, and
1159 again with the optionally the exact case.
1161 RMS had an eight level limit on directory depths from any rooted logical
1162 (allowing 16 levels overall) prior to VMS 7.2, and even with versions of
1163 VMS on VAX up through 7.3. Hence C<PERL_ROOT:[LIB.2.3.4.5.6.7.8]> is a
1164 valid directory specification but C<PERL_ROOT:[LIB.2.3.4.5.6.7.8.9]> is
1165 not. F<Makefile.PL> authors might have to take this into account, but at
1166 least they can refer to the former as C</PERL_ROOT/lib/2/3/4/5/6/7/8/>.
1168 Pumpkings and module integrators can easily see whether files with too many
1169 directory levels have snuck into the core by running the following in the
1170 top-level source directory:
1172 $ perl -ne "$_=~s/\s+.*//; print if scalar(split /\//) > 8;" < MANIFEST
1175 The VMS::Filespec module, which gets installed as part of the build
1176 process on VMS, is a pure Perl module that can easily be installed on
1177 non-VMS platforms and can be helpful for conversions to and from RMS
1178 native formats. It is also now the only way that you should check to
1179 see if VMS is in a case sensitive mode.
1181 What C<\n> represents depends on the type of file opened. It usually
1182 represents C<\012> but it could also be C<\015>, C<\012>, C<\015\012>,
1183 C<\000>, C<\040>, or nothing depending on the file organization and
1184 record format. The VMS::Stdio module provides access to the
1185 special fopen() requirements of files with unusual attributes on VMS.
1187 TCP/IP stacks are optional on VMS, so socket routines might not be
1188 implemented. UDP sockets may not be supported.
1190 The TCP/IP library support for all current versions of VMS is dynamically
1191 loaded if present, so even if the routines are configured, they may
1192 return a status indicating that they are not implemented.
1194 The value of C<$^O> on OpenVMS is "VMS". To determine the architecture
1195 that you are running on without resorting to loading all of C<%Config>
1196 you can examine the content of the C<@INC> array like so:
1198 if (grep(/VMS_AXP/, @INC)) {
1199 print "I'm on Alpha!\n";
1201 } elsif (grep(/VMS_VAX/, @INC)) {
1202 print "I'm on VAX!\n";
1204 } elsif (grep(/VMS_IA64/, @INC)) {
1205 print "I'm on IA64!\n";
1208 print "I'm not so sure about where $^O is...\n";
1211 In general, the significant differences should only be if Perl is running
1212 on VMS_VAX or one of the 64 bit OpenVMS platforms.
1214 On VMS, perl determines the UTC offset from the C<SYS$TIMEZONE_DIFFERENTIAL>
1215 logical name. Although the VMS epoch began at 17-NOV-1858 00:00:00.00,
1216 calls to C<localtime> are adjusted to count offsets from
1217 01-JAN-1970 00:00:00.00, just like Unix.
1225 F<README.vms> (installed as L<README_vms>), L<perlvms>
1229 vmsperl list, vmsperl-subscribe@perl.org
1233 vmsperl on the web, http://www.sidhe.org/vmsperl/index.html
1239 Perl on VOS is discussed in F<README.vos> in the perl distribution
1240 (installed as L<perlvos>). Perl on VOS can accept either VOS- or
1241 Unix-style file specifications as in either of the following:
1243 C<< $ perl -ne "print if /perl_setup/i" >system>notices >>
1244 C<< $ perl -ne "print if /perl_setup/i" /system/notices >>
1246 or even a mixture of both as in:
1248 C<< $ perl -ne "print if /perl_setup/i" >system/notices >>
1250 Even though VOS allows the slash character to appear in object
1251 names, because the VOS port of Perl interprets it as a pathname
1252 delimiting character, VOS files, directories, or links whose names
1253 contain a slash character cannot be processed. Such files must be
1254 renamed before they can be processed by Perl. Note that VOS limits
1255 file names to 32 or fewer characters.
1257 The value of C<$^O> on VOS is "VOS". To determine the architecture that
1258 you are running on without resorting to loading all of C<%Config> you
1259 can examine the content of the @INC array like so:
1262 print "I'm on a Stratus box!\n";
1264 print "I'm not on a Stratus box!\n";
1274 F<README.vos> (installed as L<perlvos>)
1278 The VOS mailing list.
1280 There is no specific mailing list for Perl on VOS. You can post
1281 comments to the comp.sys.stratus newsgroup, or subscribe to the general
1282 Stratus mailing list. Send a letter with "subscribe Info-Stratus" in
1283 the message body to majordomo@list.stratagy.com.
1287 VOS Perl on the web at http://ftp.stratus.com/pub/vos/posix/posix.html
1291 =head2 EBCDIC Platforms
1293 Recent versions of Perl have been ported to platforms such as OS/400 on
1294 AS/400 minicomputers as well as OS/390, VM/ESA, and BS2000 for S/390
1295 Mainframes. Such computers use EBCDIC character sets internally (usually
1296 Character Code Set ID 0037 for OS/400 and either 1047 or POSIX-BC for S/390
1297 systems). On the mainframe perl currently works under the "Unix system
1298 services for OS/390" (formerly known as OpenEdition), VM/ESA OpenEdition, or
1299 the BS200 POSIX-BC system (BS2000 is supported in perl 5.6 and greater).
1300 See L<perlos390> for details. Note that for OS/400 there is also a port of
1301 Perl 5.8.1/5.9.0 or later to the PASE which is ASCII-based (as opposed to
1302 ILE which is EBCDIC-based), see L<perlos400>.
1304 As of R2.5 of USS for OS/390 and Version 2.3 of VM/ESA these Unix
1305 sub-systems do not support the C<#!> shebang trick for script invocation.
1306 Hence, on OS/390 and VM/ESA perl scripts can be executed with a header
1307 similar to the following simple script:
1310 eval 'exec /usr/local/bin/perl -S $0 ${1+"$@"}'
1312 #!/usr/local/bin/perl # just a comment really
1314 print "Hello from perl!\n";
1316 OS/390 will support the C<#!> shebang trick in release 2.8 and beyond.
1317 Calls to C<system> and backticks can use POSIX shell syntax on all
1320 On the AS/400, if PERL5 is in your library list, you may need
1321 to wrap your perl scripts in a CL procedure to invoke them like so:
1324 CALL PGM(PERL5/PERL) PARM('/QOpenSys/hello.pl')
1327 This will invoke the perl script F<hello.pl> in the root of the
1328 QOpenSys file system. On the AS/400 calls to C<system> or backticks
1331 On these platforms, bear in mind that the EBCDIC character set may have
1332 an effect on what happens with some perl functions (such as C<chr>,
1333 C<pack>, C<print>, C<printf>, C<ord>, C<sort>, C<sprintf>, C<unpack>), as
1334 well as bit-fiddling with ASCII constants using operators like C<^>, C<&>
1335 and C<|>, not to mention dealing with socket interfaces to ASCII computers
1336 (see L<"Newlines">).
1338 Fortunately, most web servers for the mainframe will correctly
1339 translate the C<\n> in the following statement to its ASCII equivalent
1340 (C<\r> is the same under both Unix and OS/390 & VM/ESA):
1342 print "Content-type: text/html\r\n\r\n";
1344 The values of C<$^O> on some of these platforms includes:
1346 uname $^O $Config{'archname'}
1347 --------------------------------------------
1350 POSIX-BC posix-bc BS2000-posix-bc
1353 Some simple tricks for determining if you are running on an EBCDIC
1354 platform could include any of the following (perhaps all):
1356 if ("\t" eq "\05") { print "EBCDIC may be spoken here!\n"; }
1358 if (ord('A') == 193) { print "EBCDIC may be spoken here!\n"; }
1360 if (chr(169) eq 'z') { print "EBCDIC may be spoken here!\n"; }
1362 One thing you may not want to rely on is the EBCDIC encoding
1363 of punctuation characters since these may differ from code page to code
1364 page (and once your module or script is rumoured to work with EBCDIC,
1365 folks will want it to work with all EBCDIC character sets).
1373 L<perlos390>, F<README.os390>, F<perlbs2000>, F<README.vmesa>,
1378 The perl-mvs@perl.org list is for discussion of porting issues as well as
1379 general usage issues for all EBCDIC Perls. Send a message body of
1380 "subscribe perl-mvs" to majordomo@perl.org.
1384 AS/400 Perl information at
1385 http://as400.rochester.ibm.com/
1386 as well as on CPAN in the F<ports/> directory.
1390 =head2 Acorn RISC OS
1392 Because Acorns use ASCII with newlines (C<\n>) in text files as C<\012> like
1393 Unix, and because Unix filename emulation is turned on by default,
1394 most simple scripts will probably work "out of the box". The native
1395 filesystem is modular, and individual filesystems are free to be
1396 case-sensitive or insensitive, and are usually case-preserving. Some
1397 native filesystems have name length limits, which file and directory
1398 names are silently truncated to fit. Scripts should be aware that the
1399 standard filesystem currently has a name length limit of B<10>
1400 characters, with up to 77 items in a directory, but other filesystems
1401 may not impose such limitations.
1403 Native filenames are of the form
1405 Filesystem#Special_Field::DiskName.$.Directory.Directory.File
1409 Special_Field is not usually present, but may contain . and $ .
1410 Filesystem =~ m|[A-Za-z0-9_]|
1411 DsicName =~ m|[A-Za-z0-9_/]|
1412 $ represents the root directory
1413 . is the path separator
1414 @ is the current directory (per filesystem but machine global)
1415 ^ is the parent directory
1416 Directory and File =~ m|[^\0- "\.\$\%\&:\@\\^\|\177]+|
1418 The default filename translation is roughly C<tr|/.|./|;>
1420 Note that C<"ADFS::HardDisk.$.File" ne 'ADFS::HardDisk.$.File'> and that
1421 the second stage of C<$> interpolation in regular expressions will fall
1422 foul of the C<$.> if scripts are not careful.
1424 Logical paths specified by system variables containing comma-separated
1425 search lists are also allowed; hence C<System:Modules> is a valid
1426 filename, and the filesystem will prefix C<Modules> with each section of
1427 C<System$Path> until a name is made that points to an object on disk.
1428 Writing to a new file C<System:Modules> would be allowed only if
1429 C<System$Path> contains a single item list. The filesystem will also
1430 expand system variables in filenames if enclosed in angle brackets, so
1431 C<< <System$Dir>.Modules >> would look for the file
1432 S<C<$ENV{'System$Dir'} . 'Modules'>>. The obvious implication of this is
1433 that B<fully qualified filenames can start with C<< <> >>> and should
1434 be protected when C<open> is used for input.
1436 Because C<.> was in use as a directory separator and filenames could not
1437 be assumed to be unique after 10 characters, Acorn implemented the C
1438 compiler to strip the trailing C<.c> C<.h> C<.s> and C<.o> suffix from
1439 filenames specified in source code and store the respective files in
1440 subdirectories named after the suffix. Hence files are translated:
1443 C:foo.h C:h.foo (logical path variable)
1444 sys/os.h sys.h.os (C compiler groks Unix-speak)
1445 10charname.c c.10charname
1446 10charname.o o.10charname
1447 11charname_.c c.11charname (assuming filesystem truncates at 10)
1449 The Unix emulation library's translation of filenames to native assumes
1450 that this sort of translation is required, and it allows a user-defined list
1451 of known suffixes that it will transpose in this fashion. This may
1452 seem transparent, but consider that with these rules C<foo/bar/baz.h>
1453 and C<foo/bar/h/baz> both map to C<foo.bar.h.baz>, and that C<readdir> and
1454 C<glob> cannot and do not attempt to emulate the reverse mapping. Other
1455 C<.>'s in filenames are translated to C</>.
1457 As implied above, the environment accessed through C<%ENV> is global, and
1458 the convention is that program specific environment variables are of the
1459 form C<Program$Name>. Each filesystem maintains a current directory,
1460 and the current filesystem's current directory is the B<global> current
1461 directory. Consequently, sociable programs don't change the current
1462 directory but rely on full pathnames, and programs (and Makefiles) cannot
1463 assume that they can spawn a child process which can change the current
1464 directory without affecting its parent (and everyone else for that
1467 Because native operating system filehandles are global and are currently
1468 allocated down from 255, with 0 being a reserved value, the Unix emulation
1469 library emulates Unix filehandles. Consequently, you can't rely on
1470 passing C<STDIN>, C<STDOUT>, or C<STDERR> to your children.
1472 The desire of users to express filenames of the form
1473 C<< <Foo$Dir>.Bar >> on the command line unquoted causes problems,
1474 too: C<``> command output capture has to perform a guessing game. It
1475 assumes that a string C<< <[^<>]+\$[^<>]> >> is a
1476 reference to an environment variable, whereas anything else involving
1477 C<< < >> or C<< > >> is redirection, and generally manages to be 99%
1478 right. Of course, the problem remains that scripts cannot rely on any
1479 Unix tools being available, or that any tools found have Unix-like command
1482 Extensions and XS are, in theory, buildable by anyone using free
1483 tools. In practice, many don't, as users of the Acorn platform are
1484 used to binary distributions. MakeMaker does run, but no available
1485 make currently copes with MakeMaker's makefiles; even if and when
1486 this should be fixed, the lack of a Unix-like shell will cause
1487 problems with makefile rules, especially lines of the form C<cd
1488 sdbm && make all>, and anything using quoting.
1490 "S<RISC OS>" is the proper name for the operating system, but the value
1491 in C<$^O> is "riscos" (because we don't like shouting).
1495 Perl has been ported to many platforms that do not fit into any of
1496 the categories listed above. Some, such as AmigaOS, Atari MiNT,
1497 BeOS, HP MPE/iX, QNX, Plan 9, and VOS, have been well-integrated
1498 into the standard Perl source code kit. You may need to see the
1499 F<ports/> directory on CPAN for information, and possibly binaries,
1500 for the likes of: aos, Atari ST, lynxos, riscos, Novell Netware,
1501 Tandem Guardian, I<etc.> (Yes, we know that some of these OSes may
1502 fall under the Unix category, but we are not a standards body.)
1504 Some approximate operating system names and their C<$^O> values
1505 in the "OTHER" category include:
1507 OS $^O $Config{'archname'}
1508 ------------------------------------------
1509 Amiga DOS amigaos m68k-amigos
1511 MPE/iX mpeix PA-RISC1.1
1519 Amiga, F<README.amiga> (installed as L<perlamiga>).
1523 Atari, F<README.mint> and Guido Flohr's web page
1524 http://stud.uni-sb.de/~gufl0000/
1528 Be OS, F<README.beos>
1532 HP 300 MPE/iX, F<README.mpeix> and Mark Bixby's web page
1533 http://www.bixby.org/mark/perlix.html
1537 A free perl5-based PERL.NLM for Novell Netware is available in
1538 precompiled binary and source code form from http://www.novell.com/
1539 as well as from CPAN.
1543 S<Plan 9>, F<README.plan9>
1547 =head1 FUNCTION IMPLEMENTATIONS
1549 Listed below are functions that are either completely unimplemented
1550 or else have been implemented differently on various platforms.
1551 Following each description will be, in parentheses, a list of
1552 platforms that the description applies to.
1554 The list may well be incomplete, or even wrong in some places. When
1555 in doubt, consult the platform-specific README files in the Perl
1556 source distribution, and any other documentation resources accompanying
1559 Be aware, moreover, that even among Unix-ish systems there are variations.
1561 For many functions, you can also query C<%Config>, exported by
1562 default from the Config module. For example, to check whether the
1563 platform has the C<lstat> call, check C<$Config{d_lstat}>. See
1564 L<Config> for a full description of available variables.
1566 =head2 Alphabetical Listing of Perl Functions
1572 C<-r>, C<-w>, and C<-x> have a limited meaning only; directories
1573 and applications are executable, and there are no uid/gid
1574 considerations. C<-o> is not supported. (S<Mac OS>)
1576 C<-r>, C<-w>, C<-x>, and C<-o> tell whether the file is accessible,
1577 which may not reflect UIC-based file protections. (VMS)
1579 C<-s> returns the size of the data fork, not the total size of data fork
1580 plus resource fork. (S<Mac OS>).
1582 C<-s> by name on an open file will return the space reserved on disk,
1583 rather than the current extent. C<-s> on an open filehandle returns the
1584 current size. (S<RISC OS>)
1586 C<-R>, C<-W>, C<-X>, C<-O> are indistinguishable from C<-r>, C<-w>,
1587 C<-x>, C<-o>. (S<Mac OS>, Win32, VMS, S<RISC OS>)
1589 C<-b>, C<-c>, C<-k>, C<-g>, C<-p>, C<-u>, C<-A> are not implemented.
1592 C<-g>, C<-k>, C<-l>, C<-p>, C<-u>, C<-A> are not particularly meaningful.
1593 (Win32, VMS, S<RISC OS>)
1595 C<-d> is true if passed a device spec without an explicit directory.
1598 C<-T> and C<-B> are implemented, but might misclassify Mac text files
1599 with foreign characters; this is the case will all platforms, but may
1600 affect S<Mac OS> often. (S<Mac OS>)
1602 C<-x> (or C<-X>) determine if a file ends in one of the executable
1603 suffixes. C<-S> is meaningless. (Win32)
1605 C<-x> (or C<-X>) determine if a file has an executable file type.
1610 Due to issues with various CPUs, math libraries, compilers, and standards,
1611 results for C<atan2()> may vary depending on any combination of the above.
1612 Perl attempts to conform to the Open Group/IEEE standards for the results
1613 returned from C<atan2()>, but cannot force the issue if the system Perl is
1614 run on does not allow it. (Tru64, HP-UX 10.20)
1616 The current version of the standards for C<atan2()> is available at
1617 L<http://www.opengroup.org/onlinepubs/009695399/functions/atan2.html>.
1621 Meaningless. (S<Mac OS>, S<RISC OS>)
1623 Reopens file and restores pointer; if function fails, underlying
1624 filehandle may be closed, or pointer may be in a different position.
1627 The value returned by C<tell> may be affected after the call, and
1628 the filehandle may be flushed. (Win32)
1632 Only limited meaning. Disabling/enabling write permission is mapped to
1633 locking/unlocking the file. (S<Mac OS>)
1635 Only good for changing "owner" read-write access, "group", and "other"
1636 bits are meaningless. (Win32)
1638 Only good for changing "owner" and "other" read-write access. (S<RISC OS>)
1640 Access permissions are mapped onto VOS access-control list changes. (VOS)
1642 The actual permissions set depend on the value of the C<CYGWIN>
1643 in the SYSTEM environment settings. (Cygwin)
1647 Not implemented. (S<Mac OS>, Win32, S<Plan 9>, S<RISC OS>)
1649 Does nothing, but won't fail. (Win32)
1651 A little funky, because VOS's notion of ownership is a little funky (VOS).
1655 Not implemented. (S<Mac OS>, Win32, VMS, S<Plan 9>, S<RISC OS>, VOS, VM/ESA)
1659 May not be available if library or source was not provided when building
1664 Not implemented. (VMS, S<Plan 9>, VOS)
1668 Not implemented. (VMS, S<Plan 9>, VOS)
1672 Not useful. (S<Mac OS>, S<RISC OS>)
1674 Not supported. (Cygwin, Win32)
1676 Invokes VMS debugger. (VMS)
1680 Not implemented. (S<Mac OS>)
1682 Implemented via Spawn. (VM/ESA)
1684 Does not automatically flush output handles on some platforms.
1685 (SunOS, Solaris, HP-UX)
1689 Emulates UNIX exit() (which considers C<exit 1> to indicate an error) by
1690 mapping the C<1> to SS$_ABORT (C<44>). This behavior may be overridden
1691 with the pragma C<use vmsish 'exit'>. As with the CRTL's exit()
1692 function, C<exit 0> is also mapped to an exit status of SS$_NORMAL
1693 (C<1>); this mapping cannot be overridden. Any other argument to exit()
1694 is used directly as Perl's exit status. On VMS, unless the future
1695 POSIX_EXIT mode is enabled, the exit code should always be a valid
1696 VMS exit code and not a generic number. When the POSIX_EXIT mode is
1697 enabled, a generic number will be encoded in a method compatible with
1698 the C library _POSIX_EXIT macro so that it can be decoded by other
1699 programs, particularly ones written in C, like the GNV package. (VMS)
1703 Not implemented. (Win32)
1704 Some functions available based on the version of VMS. (VMS)
1708 Not implemented (S<Mac OS>, VMS, S<RISC OS>, VOS).
1710 Available only on Windows NT (not on Windows 95). (Win32)
1714 Not implemented. (S<Mac OS>, AmigaOS, S<RISC OS>, VM/ESA, VMS)
1716 Emulated using multiple interpreters. See L<perlfork>. (Win32)
1718 Does not automatically flush output handles on some platforms.
1719 (SunOS, Solaris, HP-UX)
1723 Not implemented. (S<Mac OS>, S<RISC OS>)
1727 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>)
1731 Not implemented. (S<Mac OS>, Win32, S<RISC OS>)
1735 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS, VM/ESA)
1739 Not implemented. (S<Mac OS>, Win32)
1741 Not useful. (S<RISC OS>)
1745 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>)
1749 Not implemented. (S<Mac OS>, Win32, S<Plan 9>)
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>)
1765 =item getprotobynumber
1767 Not implemented. (S<Mac OS>)
1771 Not implemented. (S<Mac OS>)
1775 Not implemented. (S<Mac OS>, Win32, VM/ESA)
1779 Not implemented. (S<Mac OS>, Win32, VMS, VM/ESA)
1783 C<gethostbyname('localhost')> does not work everywhere: you may have
1784 to use C<gethostbyname('127.0.0.1')>. (S<Mac OS>, S<Irix 5>)
1788 Not implemented. (S<Mac OS>, Win32)
1792 Not implemented. (S<Mac OS>, Win32, S<Plan 9>)
1796 Not implemented. (S<Mac OS>, Win32, S<Plan 9>)
1800 Not implemented. (Win32, S<Plan 9>)
1804 Not implemented. (S<Mac OS>, Win32, S<Plan 9>, S<RISC OS>)
1808 Not implemented. (S<Mac OS>, Win32, S<Plan 9>, S<RISC OS>)
1812 Not implemented. (S<Mac OS>, Win32, S<Plan 9>, S<RISC OS>)
1816 Not implemented. (S<Plan 9>, Win32, S<RISC OS>)
1820 Not implemented. (S<Mac OS>, MPE/iX, VM/ESA, Win32)
1824 Not implemented. (S<Mac OS>, MPE/iX, S<RISC OS>, VM/ESA, VMS, Win32)
1828 Not implemented. (S<Mac OS>, Win32)
1832 Not implemented. (S<Mac OS>, Win32, S<Plan 9>)
1836 Not implemented. (S<Mac OS>, Win32, S<Plan 9>)
1840 Not implemented. (S<Plan 9>, Win32)
1842 =item getsockopt SOCKET,LEVEL,OPTNAME
1844 Not implemented. (S<Plan 9>)
1848 This operator is implemented via the File::Glob extension on most
1849 platforms. See L<File::Glob> for portability information.
1853 Same portability caveats as L<localtime>.
1855 =item ioctl FILEHANDLE,FUNCTION,SCALAR
1857 Not implemented. (VMS)
1859 Available only for socket handles, and it does what the ioctlsocket() call
1860 in the Winsock API does. (Win32)
1862 Available only for socket handles. (S<RISC OS>)
1866 C<kill(0, LIST)> is implemented for the sake of taint checking;
1867 use with other signals is unimplemented. (S<Mac OS>)
1869 Not implemented, hence not useful for taint checking. (S<RISC OS>)
1871 C<kill()> doesn't have the semantics of C<raise()>, i.e. it doesn't send
1872 a signal to the identified process like it does on Unix platforms.
1873 Instead C<kill($sig, $pid)> terminates the process identified by $pid,
1874 and makes it exit immediately with exit status $sig. As in Unix, if
1875 $sig is 0 and the specified process exists, it returns true without
1876 actually terminating it. (Win32)
1878 Is not supported for process identification number of 0 or negative
1883 Not implemented. (S<Mac OS>, MPE/iX, S<RISC OS>)
1885 Link count not updated because hard links are not quite that hard
1886 (They are sort of half-way between hard and soft links). (AmigaOS)
1888 Hard links are implemented on Win32 under NTFS only. They are
1889 natively supported on Windows 2000 and later. On Windows NT they
1890 are implemented using the Windows POSIX subsystem support and the
1891 Perl process will need Administrator or Backup Operator privileges
1892 to create hard links.
1894 Available on 64 bit OpenVMS 8.2 and later. (VMS)
1898 Because Perl currently relies on the native standard C localtime()
1899 function, it is only safe to use times between 0 and (2**31)-1. Times
1900 outside this range may result in unexpected behavior depending on your
1901 operating system's implementation of localtime().
1905 Not implemented. (S<RISC OS>)
1907 Return values (especially for device and inode) may be bogus. (Win32)
1917 Not implemented. (S<Mac OS>, Win32, VMS, S<Plan 9>, S<RISC OS>, VOS)
1921 The C<|> variants are supported only if ToolServer is installed.
1924 open to C<|-> and C<-|> are unsupported. (S<Mac OS>, Win32, S<RISC OS>)
1926 Opening a process does not automatically flush output handles on some
1927 platforms. (SunOS, Solaris, HP-UX)
1931 Very limited functionality. (MiNT)
1935 Not implemented. (Win32, VMS, S<RISC OS>)
1939 Can't move directories between directories on different logical volumes. (Win32)
1943 Only implemented on sockets. (Win32, VMS)
1945 Only reliable on sockets. (S<RISC OS>)
1947 Note that the C<select FILEHANDLE> form is generally portable.
1955 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS)
1959 Not implemented. (S<Mac OS>, MPE/iX, VMS, Win32, S<RISC OS>, VOS)
1963 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS)
1967 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS)
1971 Not implemented. (S<Mac OS>, MPE/iX, Win32, S<RISC OS>, VOS)
1975 Not implemented. (S<Plan 9>)
1985 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS)
1989 A relatively recent addition to socket functions, may not
1990 be implemented even in UNIX platforms.
1994 Not implemented. (Win32, S<RISC OS>, VOS, VM/ESA)
1996 Available on 64 bit OpenVMS 8.2 and later. (VMS)
2000 Platforms that do not have rdev, blksize, or blocks will return these
2001 as '', so numeric comparison or manipulation of these fields may cause
2002 'not numeric' warnings.
2004 mtime and atime are the same thing, and ctime is creation time instead of
2005 inode change time. (S<Mac OS>).
2007 ctime not supported on UFS (S<Mac OS X>).
2009 ctime is creation time instead of inode change time (Win32).
2011 device and inode are not meaningful. (Win32)
2013 device and inode are not necessarily reliable. (VMS)
2015 mtime, atime and ctime all return the last modification time. Device and
2016 inode are not necessarily reliable. (S<RISC OS>)
2018 dev, rdev, blksize, and blocks are not available. inode is not
2019 meaningful and will differ between stat calls on the same file. (os2)
2021 some versions of cygwin when doing a stat("foo") and if not finding it
2022 may then attempt to stat("foo.exe") (Cygwin)
2024 On Win32 stat() needs to open the file to determine the link count
2025 and update attributes that may have been changed through hard links.
2026 Setting ${^WIN32_SLOPPY_STAT} to a true value speeds up stat() by
2027 not performing this operation. (Win32)
2031 Not implemented. (Win32, VMS, S<RISC OS>)
2035 Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS, VM/ESA)
2039 The traditional "0", "1", and "2" MODEs are implemented with different
2040 numeric values on some systems. The flags exported by C<Fcntl>
2041 (O_RDONLY, O_WRONLY, O_RDWR) should work everywhere though. (S<Mac
2042 OS>, OS/390, VM/ESA)
2046 Only implemented if ToolServer is installed. (S<Mac OS>)
2048 As an optimization, may not call the command shell specified in
2049 C<$ENV{PERL5SHELL}>. C<system(1, @args)> spawns an external
2050 process and immediately returns its process designator, without
2051 waiting for it to terminate. Return value may be used subsequently
2052 in C<wait> or C<waitpid>. Failure to spawn() a subprocess is indicated
2053 by setting $? to "255 << 8". C<$?> is set in a way compatible with
2054 Unix (i.e. the exitstatus of the subprocess is obtained by "$? >> 8",
2055 as described in the documentation). (Win32)
2057 There is no shell to process metacharacters, and the native standard is
2058 to pass a command line terminated by "\n" "\r" or "\0" to the spawned
2059 program. Redirection such as C<< > foo >> is performed (if at all) by
2060 the run time library of the spawned program. C<system> I<list> will call
2061 the Unix emulation library's C<exec> emulation, which attempts to provide
2062 emulation of the stdin, stdout, stderr in force in the parent, providing
2063 the child program uses a compatible version of the emulation library.
2064 I<scalar> will call the native command line direct and no such emulation
2065 of a child Unix program will exists. Mileage B<will> vary. (S<RISC OS>)
2067 Far from being POSIX compliant. Because there may be no underlying
2068 /bin/sh tries to work around the problem by forking and execing the
2069 first token in its argument string. Handles basic redirection
2070 ("<" or ">") on its own behalf. (MiNT)
2072 Does not automatically flush output handles on some platforms.
2073 (SunOS, Solaris, HP-UX)
2075 The return value is POSIX-like (shifted up by 8 bits), which only allows
2076 room for a made-up value derived from the severity bits of the native
2077 32-bit condition code (unless overridden by C<use vmsish 'status'>).
2078 If the native condition code is one that has a POSIX value encoded, the
2079 POSIX value will be decoded to extract the expected exit value.
2080 For more details see L<perlvms/$?>. (VMS)
2084 Only the first entry returned is nonzero. (S<Mac OS>)
2086 "cumulative" times will be bogus. On anything other than Windows NT
2087 or Windows 2000, "system" time will be bogus, and "user" time is
2088 actually the time returned by the clock() function in the C runtime
2091 Not useful. (S<RISC OS>)
2095 Not implemented. (Older versions of VMS)
2097 Truncation to same-or-shorter lengths only. (VOS)
2099 If a FILEHANDLE is supplied, it must be writable and opened in append
2100 mode (i.e., use C<<< open(FH, '>>filename') >>>
2101 or C<sysopen(FH,...,O_APPEND|O_RDWR)>. If a filename is supplied, it
2102 should not be held open elsewhere. (Win32)
2106 Returns undef where unavailable, as of version 5.005.
2108 C<umask> works but the correct permissions are set only when the file
2109 is finally closed. (AmigaOS)
2113 Only the modification time is updated. (S<BeOS>, S<Mac OS>, VMS, S<RISC OS>)
2115 May not behave as expected. Behavior depends on the C runtime
2116 library's implementation of utime(), and the filesystem being
2117 used. The FAT filesystem typically does not support an "access
2118 time" field, and it may limit timestamps to a granularity of
2119 two seconds. (Win32)
2125 Not implemented. (S<Mac OS>)
2127 Can only be applied to process handles returned for processes spawned
2128 using C<system(1, ...)> or pseudo processes created with C<fork()>. (Win32)
2130 Not useful. (S<RISC OS>)
2135 =head1 Supported Platforms
2137 As of July 2002 (the Perl release 5.8.0), the following platforms are
2138 able to build Perl from the standard source code distribution
2139 available at http://www.cpan.org/src/index.html
2150 HI-UXMPP (Hitachi) (5.8.0 worked but we didn't know it)
2160 ReliantUNIX (formerly SINIX)
2162 OpenVMS (formerly VMS)
2163 Open UNIX (Unixware) (since Perl 5.8.1/5.9.0)
2165 OS/400 (using the PASE) (since Perl 5.8.1/5.9.0)
2167 POSIX-BC (formerly BS2000)
2172 Tru64 UNIX (formerly DEC OSF/1, Digital UNIX)
2177 Win95/98/ME/2K/XP 2)
2179 z/OS (formerly OS/390)
2182 1) in DOS mode either the DOS or OS/2 ports can be used
2183 2) compilers: Borland, MinGW (GCC), VC6
2185 The following platforms worked with the previous releases (5.6 and
2186 5.7), but we did not manage either to fix or to test these in time
2187 for the 5.8.0 release. There is a very good chance that many of these
2188 will work fine with the 5.8.0.
2201 Known to be broken for 5.8.0 (but 5.6.1 and 5.7.2 can be used):
2205 The following platforms have been known to build Perl from source in
2206 the past (5.005_03 and earlier), but we haven't been able to verify
2207 their status for the current release, either because the
2208 hardware/software platforms are rare or because we don't have an
2209 active champion on these platforms--or both. They used to work,
2210 though, so go ahead and try compiling them, and let perlbug@perl.org
2244 The following platforms have their own source code distributions and
2245 binaries available via http://www.cpan.org/ports/
2249 OS/400 (ILE) 5.005_02
2250 Tandem Guardian 5.004
2252 The following platforms have only binaries available via
2253 http://www.cpan.org/ports/index.html :
2257 Acorn RISCOS 5.005_02
2261 Although we do suggest that you always build your own Perl from
2262 the source code, both for maximal configurability and for security,
2263 in case you are in a hurry you can check
2264 http://www.cpan.org/ports/index.html for binary distributions.
2268 L<perlaix>, L<perlamiga>, L<perlapollo>, L<perlbeos>, L<perlbs2000>,
2269 L<perlce>, L<perlcygwin>, L<perldgux>, L<perldos>, L<perlepoc>,
2270 L<perlebcdic>, L<perlfreebsd>, L<perlhurd>, L<perlhpux>, L<perlirix>,
2271 L<perlmachten>, L<perlmacos>, L<perlmacosx>, L<perlmint>, L<perlmpeix>,
2272 L<perlnetware>, L<perlos2>, L<perlos390>, L<perlos400>,
2273 L<perlplan9>, L<perlqnx>, L<perlsolaris>, L<perltru64>,
2274 L<perlunicode>, L<perlvmesa>, L<perlvms>, L<perlvos>,
2275 L<perlwin32>, and L<Win32>.
2277 =head1 AUTHORS / CONTRIBUTORS
2279 Abigail <abigail@foad.org>,
2280 Charles Bailey <bailey@newman.upenn.edu>,
2281 Graham Barr <gbarr@pobox.com>,
2282 Tom Christiansen <tchrist@perl.com>,
2283 Nicholas Clark <nick@ccl4.org>,
2284 Thomas Dorner <Thomas.Dorner@start.de>,
2285 Andy Dougherty <doughera@lafayette.edu>,
2286 Dominic Dunlop <domo@computer.org>,
2287 Neale Ferguson <neale@vma.tabnsw.com.au>,
2288 David J. Fiander <davidf@mks.com>,
2289 Paul Green <Paul.Green@stratus.com>,
2290 M.J.T. Guy <mjtg@cam.ac.uk>,
2291 Jarkko Hietaniemi <jhi@iki.fi>,
2292 Luther Huffman <lutherh@stratcom.com>,
2293 Nick Ing-Simmons <nick@ing-simmons.net>,
2294 Andreas J. KE<ouml>nig <a.koenig@mind.de>,
2295 Markus Laker <mlaker@contax.co.uk>,
2296 Andrew M. Langmead <aml@world.std.com>,
2297 Larry Moore <ljmoore@freespace.net>,
2298 Paul Moore <Paul.Moore@uk.origin-it.com>,
2299 Chris Nandor <pudge@pobox.com>,
2300 Matthias Neeracher <neeracher@mac.com>,
2301 Philip Newton <pne@cpan.org>,
2302 Gary Ng <71564.1743@CompuServe.COM>,
2303 Tom Phoenix <rootbeer@teleport.com>,
2304 AndrE<eacute> Pirard <A.Pirard@ulg.ac.be>,
2305 Peter Prymmer <pvhp@forte.com>,
2306 Hugo van der Sanden <hv@crypt0.demon.co.uk>,
2307 Gurusamy Sarathy <gsar@activestate.com>,
2308 Paul J. Schinder <schinder@pobox.com>,
2309 Michael G Schwern <schwern@pobox.com>,
2310 Dan Sugalski <dan@sidhe.org>,
2311 Nathan Torkington <gnat@frii.com>.
2312 John Malmberg <wb8tyw@qsl.net>