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 has 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 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 S<Mac OS> uses C<\015>, and z/OS uses C<\025>.
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>. On EBCDIC platforms, C<\n> could be C<\025> or C<\045>.
92 In DOSish perls, C<\n> usually means C<\012>, but when
93 accessing a file in "text" mode, perl uses the C<:crlf> layer that
94 translates it to (or from) C<\015\012>, depending on whether you're
95 reading or writing. Unix does the same thing on ttys in canonical
96 mode. C<\015\012> is commonly referred to as CRLF.
98 To trim trailing newlines from text lines use C<chomp()>. With default
99 settings that function looks for a trailing C<\n> character and thus
100 trims in a portable way.
102 When dealing with binary files (or text files in binary mode) be sure
103 to explicitly set $/ to the appropriate value for your file format
104 before using C<chomp()>.
106 Because of the "text" mode translation, DOSish perls have limitations
107 in using C<seek> and C<tell> on a file accessed in "text" mode.
108 Stick to C<seek>-ing to locations you got from C<tell> (and no
109 others), and you are usually free to use C<seek> and C<tell> even
110 in "text" mode. Using C<seek> or C<tell> or other file operations
111 may be non-portable. If you use C<binmode> on a file, however, you
112 can usually C<seek> and C<tell> with arbitrary values safely.
114 A common misconception in socket programming is that S<C<\n eq \012>>
115 everywhere. When using protocols such as common Internet protocols,
116 C<\012> and C<\015> are called for specifically, and the values of
117 the logical C<\n> and C<\r> (carriage return) are not reliable.
119 print SOCKET "Hi there, client!\r\n"; # WRONG
120 print SOCKET "Hi there, client!\015\012"; # RIGHT
122 However, using C<\015\012> (or C<\cM\cJ>, or C<\x0D\x0A>) can be tedious
123 and unsightly, as well as confusing to those maintaining the code. As
124 such, the C<Socket> module supplies the Right Thing for those who want it.
126 use Socket qw(:DEFAULT :crlf);
127 print SOCKET "Hi there, client!$CRLF" # RIGHT
129 When reading from a socket, remember that the default input record
130 separator C<$/> is C<\n>, but robust socket code will recognize as
131 either C<\012> or C<\015\012> as end of line:
133 while (<SOCKET>) { # NOT ADVISABLE!
137 Because both CRLF and LF end in LF, the input record separator can
138 be set to LF and any CR stripped later. Better to write:
140 use Socket qw(:DEFAULT :crlf);
141 local($/) = LF; # not needed if $/ is already \012
144 s/$CR?$LF/\n/; # not sure if socket uses LF or CRLF, OK
145 # s/\015?\012/\n/; # same thing
148 This example is preferred over the previous one--even for Unix
149 platforms--because now any C<\015>'s (C<\cM>'s) are stripped out
150 (and there was much rejoicing).
152 Similarly, functions that return text data--such as a function that
153 fetches a web page--should sometimes translate newlines before
154 returning the data, if they've not yet been translated to the local
155 newline representation. A single line of code will often suffice:
157 $data =~ s/\015?\012/\n/g;
160 Some of this may be confusing. Here's a handy reference to the ASCII CR
161 and LF characters. You can print it out and stick it in your wallet.
163 LF eq \012 eq \x0A eq \cJ eq chr(10) eq ASCII 10
164 CR eq \015 eq \x0D eq \cM eq chr(13) eq ASCII 13
167 ---------------------------
170 \n * | LF | CRLF | CR |
171 \r * | CR | CR | LF |
172 ---------------------------
175 The Unix column assumes that you are not accessing a serial line
176 (like a tty) in canonical mode. If you are, then CR on input becomes
177 "\n", and "\n" on output becomes CRLF.
179 These are just the most common definitions of C<\n> and C<\r> in Perl.
180 There may well be others. For example, on an EBCDIC implementation
181 such as z/OS (OS/390) or OS/400 (using the ILE, the PASE is ASCII-based)
182 the above material is similar to "Unix" but the code numbers change:
184 LF eq \025 eq \x15 eq \cU eq chr(21) eq CP-1047 21
185 LF eq \045 eq \x25 eq chr(37) eq CP-0037 37
186 CR eq \015 eq \x0D eq \cM eq chr(13) eq CP-1047 13
187 CR eq \015 eq \x0D eq \cM eq chr(13) eq CP-0037 13
190 ----------------------
195 ----------------------
198 =head2 Numbers endianness and Width
200 Different CPUs store integers and floating point numbers in different
201 orders (called I<endianness>) and widths (32-bit and 64-bit being the
202 most common today). This affects your programs when they attempt to transfer
203 numbers in binary format from one CPU architecture to another,
204 usually either "live" via network connection, or by storing the
205 numbers to secondary storage such as a disk file or tape.
207 Conflicting storage orders make an utter mess out of the numbers. If a
208 little-endian host (Intel, VAX) stores 0x12345678 (305419896 in
209 decimal), a big-endian host (Motorola, Sparc, PA) reads it as
210 0x78563412 (2018915346 in decimal). Alpha and MIPS can be either:
211 Digital/Compaq used/uses them in little-endian mode; SGI/Cray uses
212 them in big-endian mode. To avoid this problem in network (socket)
213 connections use the C<pack> and C<unpack> formats C<n> and C<N>, the
214 "network" orders. These are guaranteed to be portable.
216 As of Perl 5.10.0, you can also use the C<E<gt>> and C<E<lt>> modifiers
217 to force big- or little-endian byte-order. This is useful if you want
218 to store signed integers or 64-bit integers, for example.
220 You can explore the endianness of your platform by unpacking a
221 data structure packed in native format such as:
223 print unpack("h*", pack("s2", 1, 2)), "\n";
224 # '10002000' on e.g. Intel x86 or Alpha 21064 in little-endian mode
225 # '00100020' on e.g. Motorola 68040
227 If you need to distinguish between endian architectures you could use
228 either of the variables set like so:
230 $is_big_endian = unpack("h*", pack("s", 1)) =~ /01/;
231 $is_little_endian = unpack("h*", pack("s", 1)) =~ /^1/;
233 Differing widths can cause truncation even between platforms of equal
234 endianness. The platform of shorter width loses the upper parts of the
235 number. There is no good solution for this problem except to avoid
236 transferring or storing raw binary numbers.
238 One can circumnavigate both these problems in two ways. Either
239 transfer and store numbers always in text format, instead of raw
240 binary, or else consider using modules like C<Data::Dumper> and
242 (included as of Perl 5.8). Keeping all data as text significantly
245 =head2 Files and Filesystems
247 Most platforms these days structure files in a hierarchical fashion.
248 So, it is reasonably safe to assume that all platforms support the
249 notion of a "path" to uniquely identify a file on the system. How
250 that path is really written, though, differs considerably.
252 Although similar, file path specifications differ between Unix,
253 Windows, S<Mac OS>, OS/2, VMS, VOS, S<RISC OS>, and probably others.
254 Unix, for example, is one of the few OSes that has the elegant idea
255 of a single root directory.
257 DOS, OS/2, VMS, VOS, and Windows can work similarly to Unix with C</>
258 as path separator, or in their own idiosyncratic ways (such as having
259 several root directories and various "unrooted" device files such NIL:
262 S<Mac OS> 9 and earlier used C<:> as a path separator instead of C</>.
264 The filesystem may support neither hard links (C<link>) nor
265 symbolic links (C<symlink>, C<readlink>, C<lstat>).
267 The filesystem may support neither access timestamp nor change
268 timestamp (meaning that about the only portable timestamp is the
269 modification timestamp), or one second granularity of any timestamps
270 (e.g. the FAT filesystem limits the time granularity to two seconds).
272 The "inode change timestamp" (the C<-C> filetest) may really be the
273 "creation timestamp" (which it is not in Unix).
275 VOS perl can emulate Unix filenames with C</> as path separator. The
276 native pathname characters greater-than, less-than, number-sign, and
277 percent-sign are always accepted.
279 S<RISC OS> perl can emulate Unix filenames with C</> as path
280 separator, or go native and use C<.> for path separator and C<:> to
281 signal filesystems and disk names.
283 Don't assume Unix filesystem access semantics: that read, write,
284 and execute are all the permissions there are, and even if they exist,
285 that their semantics (for example what do C<"r">, C<"w">, and C<"x"> mean on
286 a directory) are the Unix ones. The various Unix/POSIX compatibility
287 layers usually try to make interfaces like C<chmod()> work, but sometimes
288 there simply is no good mapping.
290 The C<File::Spec> modules provide methods to manipulate path
291 specifications and return the results in native format for each
292 platform. This is often unnecessary as Unix-style paths are
293 understood by Perl on every supported platform, but if you need to
294 produce native paths for a native utility that does not understand
295 Unix syntax, or if you are operating on paths or path components
296 in unknown (and thus possibly native) syntax, C<File::Spec> is
297 your friend. Here are two brief examples:
299 use File::Spec::Functions;
300 chdir(updir()); # go up one directory
302 # Concatenate a path from its components
303 my $file = catfile(updir(), 'temp', 'file.txt');
304 # on Unix: '../temp/file.txt'
305 # on Win32: '..\temp\file.txt'
306 # on VMS: '[-.temp]file.txt'
308 In general, production code should not have file paths hardcoded.
309 Making them user-supplied or read from a configuration file is
310 better, keeping in mind that file path syntax varies on different
313 This is especially noticeable in scripts like Makefiles and test suites,
314 which often assume C</> as a path separator for subdirectories.
316 Also of use is C<File::Basename> from the standard distribution, which
317 splits a pathname into pieces (base filename, full path to directory,
320 Even when on a single platform (if you can call Unix a single platform),
321 remember not to count on the existence or the contents of particular
322 system-specific files or directories, like F</etc/passwd>,
323 F</etc/sendmail.conf>, F</etc/resolv.conf>, or even F</tmp/>. For
324 example, F</etc/passwd> may exist but not contain the encrypted
325 passwords, because the system is using some form of enhanced security.
326 Or it may not contain all the accounts, because the system is using NIS.
327 If code does need to rely on such a file, include a description of the
328 file and its format in the code's documentation, then make it easy for
329 the user to override the default location of the file.
331 Don't assume a text file will end with a newline. They should,
334 Do not have two files or directories of the same name with different
335 case, like F<test.pl> and F<Test.pl>, as many platforms have
336 case-insensitive (or at least case-forgiving) filenames. Also, try
337 not to have non-word characters (except for C<.>) in the names, and
338 keep them to the 8.3 convention, for maximum portability, onerous a
339 burden though this may appear.
341 Likewise, when using the C<AutoSplit> module, try to keep your functions to
342 8.3 naming and case-insensitive conventions; or, at the least,
343 make it so the resulting files have a unique (case-insensitively)
346 Whitespace in filenames is tolerated on most systems, but not all,
347 and even on systems where it might be tolerated, some utilities
348 might become confused by such whitespace.
350 Many systems (DOS, VMS ODS-2) cannot have more than one C<.> in their
353 Don't assume C<< > >> won't be the first character of a filename.
354 Always use C<< < >> explicitly to open a file for reading, or even
355 better, use the three-arg version of C<open>, unless you want the user to
356 be able to specify a pipe open.
358 open my $fh, '<', $existing_file) or die $!;
360 If filenames might use strange characters, it is safest to open it
361 with C<sysopen> instead of C<open>. C<open> is magic and can
362 translate characters like C<< > >>, C<< < >>, and C<|>, which may
363 be the wrong thing to do. (Sometimes, though, it's the right thing.)
364 Three-arg open can also help protect against this translation in cases
365 where it is undesirable.
367 Don't use C<:> as a part of a filename since many systems use that for
368 their own semantics (Mac OS Classic for separating pathname components,
369 many networking schemes and utilities for separating the nodename and
370 the pathname, and so on). For the same reasons, avoid C<@>, C<;> and
373 Don't assume that in pathnames you can collapse two leading slashes
374 C<//> into one: some networking and clustering filesystems have special
375 semantics for that. Let the operating system sort it out.
377 The I<portable filename characters> as defined by ANSI C are
379 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
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
384 and the C<"-"> shouldn't be the first character. If you want to be
385 hypercorrect, stay case-insensitive and within the 8.3 naming
386 convention (all the files and directories have to be unique within one
387 directory if their names are lowercased and truncated to eight
388 characters before the C<.>, if any, and to three characters after the
389 C<.>, if any). (And do not use C<.>s in directory names.)
391 =head2 System Interaction
393 Not all platforms provide a command line. These are usually platforms
394 that rely primarily on a Graphical User Interface (GUI) for user
395 interaction. A program requiring a command line interface might
396 not work everywhere. This is probably for the user of the program
397 to deal with, so don't stay up late worrying about it.
399 Some platforms can't delete or rename files held open by the system,
400 this limitation may also apply to changing filesystem metainformation
401 like file permissions or owners. Remember to C<close> files when you
402 are done with them. Don't C<unlink> or C<rename> an open file. Don't
403 C<tie> or C<open> a file already tied or opened; C<untie> or C<close>
406 Don't open the same file more than once at a time for writing, as some
407 operating systems put mandatory locks on such files.
409 Don't assume that write/modify permission on a directory gives the
410 right to add or delete files/directories in that directory. That is
411 filesystem specific: in some filesystems you need write/modify
412 permission also (or even just) in the file/directory itself. In some
413 filesystems (AFS, DFS) the permission to add/delete directory entries
414 is a completely separate permission.
416 Don't assume that a single C<unlink> completely gets rid of the file:
417 some filesystems (most notably the ones in VMS) have versioned
418 filesystems, and C<unlink()> removes only the most recent one (it doesn't
419 remove all the versions because by default the native tools on those
420 platforms remove just the most recent version, too). The portable
421 idiom to remove all the versions of a file is
423 1 while unlink "file";
425 This will terminate if the file is undeleteable for some reason
426 (protected, not there, and so on).
428 Don't count on a specific environment variable existing in C<%ENV>.
429 Don't count on C<%ENV> entries being case-sensitive, or even
430 case-preserving. Don't try to clear C<%ENV> by saying C<%ENV = ();>, or,
431 if you really have to, make it conditional on C<$^O ne 'VMS'> since in
432 VMS the C<%ENV> table is much more than a per-process key-value string
435 On VMS, some entries in the C<%ENV> hash are dynamically created when
436 their key is used on a read if they did not previously exist. The
437 values for C<$ENV{HOME}>, C<$ENV{TERM}>, C<$ENV{PATH}>, and C<$ENV{USER}>,
438 are known to be dynamically generated. The specific names that are
439 dynamically generated may vary with the version of the C library on VMS,
440 and more may exist than are documented.
442 On VMS by default, changes to the %ENV hash persist after perl exits.
443 Subsequent invocations of perl in the same process can inadvertently
444 inherit environment settings that were meant to be temporary.
446 Don't count on signals or C<%SIG> for anything.
448 Don't count on filename globbing. Use C<opendir>, C<readdir>, and
451 Don't count on per-program environment variables, or per-program current
454 Don't count on specific values of C<$!>, neither numeric nor
455 especially the string values. Users may switch their locales causing
456 error messages to be translated into their languages. If you can
457 trust a POSIXish environment, you can portably use the symbols defined
458 by the C<Errno> module, like C<ENOENT>. And don't trust on the values of C<$!>
459 at all except immediately after a failed system call.
461 =head2 Command names versus file pathnames
463 Don't assume that the name used to invoke a command or program with
464 C<system> or C<exec> can also be used to test for the existence of the
465 file that holds the executable code for that command or program.
466 First, many systems have "internal" commands that are built-in to the
467 shell or OS and while these commands can be invoked, there is no
468 corresponding file. Second, some operating systems (e.g., Cygwin,
469 DJGPP, OS/2, and VOS) have required suffixes for executable files;
470 these suffixes are generally permitted on the command name but are not
471 required. Thus, a command like F<"perl"> might exist in a file named
472 F<"perl">, F<"perl.exe">, or F<"perl.pm">, depending on the operating system.
473 The variable C<"_exe"> in the C<Config> module holds the executable suffix,
474 if any. Third, the VMS port carefully sets up C<$^X> and
475 C<$Config{perlpath}> so that no further processing is required. This is
476 just as well, because the matching regular expression used below would
477 then have to deal with a possible trailing version number in the VMS
480 To convert C<$^X> to a file pathname, taking account of the requirements
481 of the various operating system possibilities, say:
486 {$thisperl .= $Config{_exe} unless $thisperl =~ m/$Config{_exe}$/i;}
488 To convert C<$Config{perlpath}> to a file pathname, say:
491 my $thisperl = $Config{perlpath};
493 {$thisperl .= $Config{_exe} unless $thisperl =~ m/$Config{_exe}$/i;}
497 Don't assume that you can reach the public Internet.
499 Don't assume that there is only one way to get through firewalls
500 to the public Internet.
502 Don't assume that you can reach outside world through any other port
503 than 80, or some web proxy. ftp is blocked by many firewalls.
505 Don't assume that you can send email by connecting to the local SMTP port.
507 Don't assume that you can reach yourself or any node by the name
508 'localhost'. The same goes for '127.0.0.1'. You will have to try both.
510 Don't assume that the host has only one network card, or that it
511 can't bind to many virtual IP addresses.
513 Don't assume a particular network device name.
515 Don't assume a particular set of C<ioctl()>s will work.
517 Don't assume that you can ping hosts and get replies.
519 Don't assume that any particular port (service) will respond.
521 Don't assume that C<Sys::Hostname> (or any other API or command) returns
522 either a fully qualified hostname or a non-qualified hostname: it all
523 depends on how the system had been configured. Also remember that for
524 things such as DHCP and NAT, the hostname you get back might not be
527 All the above "don't":s may look daunting, and they are, but the key
528 is to degrade gracefully if one cannot reach the particular network
529 service one wants. Croaking or hanging do not look very professional.
531 =head2 Interprocess Communication (IPC)
533 In general, don't directly access the system in code meant to be
534 portable. That means, no C<system>, C<exec>, C<fork>, C<pipe>,
535 C<``>, C<qx//>, C<open> with a C<|>, nor any of the other things
536 that makes being a Perl hacker worth being.
538 Commands that launch external processes are generally supported on
539 most platforms (though many of them do not support any type of
540 forking). The problem with using them arises from what you invoke
541 them on. External tools are often named differently on different
542 platforms, may not be available in the same location, might accept
543 different arguments, can behave differently, and often present their
544 results in a platform-dependent way. Thus, you should seldom depend
545 on them to produce consistent results. (Then again, if you're calling
546 I<netstat -a>, you probably don't expect it to run on both Unix and CP/M.)
548 One especially common bit of Perl code is opening a pipe to B<sendmail>:
550 open(MAIL, '|/usr/lib/sendmail -t')
551 or die "cannot fork sendmail: $!";
553 This is fine for systems programming when sendmail is known to be
554 available. But it is not fine for many non-Unix systems, and even
555 some Unix systems that may not have sendmail installed. If a portable
556 solution is needed, see the various distributions on CPAN that deal
557 with it. C<Mail::Mailer> and C<Mail::Send> in the C<MailTools> distribution are
558 commonly used, and provide several mailing methods, including C<mail>,
559 C<sendmail>, and direct SMTP (via C<Net::SMTP>) if a mail transfer agent is
560 not available. C<Mail::Sendmail> is a standalone module that provides
561 simple, platform-independent mailing.
563 The Unix System V IPC (C<msg*(), sem*(), shm*()>) is not available
564 even on all Unix platforms.
566 Do not use either the bare result of C<pack("N", 10, 20, 30, 40)> or
567 bare v-strings (such as C<v10.20.30.40>) to represent IPv4 addresses:
568 both forms just pack the four bytes into network order. That this
569 would be equal to the C language C<in_addr> struct (which is what the
570 socket code internally uses) is not guaranteed. To be portable use
571 the routines of the C<Socket> extension, such as C<inet_aton()>,
572 C<inet_ntoa()>, and C<sockaddr_in()>.
574 The rule of thumb for portable code is: Do it all in portable Perl, or
575 use a module (that may internally implement it with platform-specific
576 code, but exposes a common interface).
578 =head2 External Subroutines (XS)
580 XS code can usually be made to work with any platform, but dependent
581 libraries, header files, etc., might not be readily available or
582 portable, or the XS code itself might be platform-specific, just as Perl
583 code might be. If the libraries and headers are portable, then it is
584 normally reasonable to make sure the XS code is portable, too.
586 A different type of portability issue arises when writing XS code:
587 availability of a C compiler on the end-user's system. C brings
588 with it its own portability issues, and writing XS code will expose
589 you to some of those. Writing purely in Perl is an easier way to
592 =head2 Standard Modules
594 In general, the standard modules work across platforms. Notable
595 exceptions are the C<CPAN> module (which currently makes connections to external
596 programs that may not be available), platform-specific modules (like
597 C<ExtUtils::MM_VMS>), and DBM modules.
599 There is no one DBM module available on all platforms.
600 C<SDBM_File> and the others are generally available on all Unix and DOSish
601 ports, but not in MacPerl, where only C<NDBM_File> and C<DB_File> are
604 The good news is that at least some DBM module should be available, and
605 C<AnyDBM_File> will use whichever module it can find. Of course, then
606 the code needs to be fairly strict, dropping to the greatest common
607 factor (e.g., not exceeding 1K for each record), so that it will
608 work with any DBM module. See L<AnyDBM_File> for more details.
612 The system's notion of time of day and calendar date is controlled in
613 widely different ways. Don't assume the timezone is stored in C<$ENV{TZ}>,
614 and even if it is, don't assume that you can control the timezone through
615 that variable. Don't assume anything about the three-letter timezone
616 abbreviations (for example that MST would be the Mountain Standard Time,
617 it's been known to stand for Moscow Standard Time). If you need to
618 use timezones, express them in some unambiguous format like the
619 exact number of minutes offset from UTC, or the POSIX timezone
622 Don't assume that the epoch starts at 00:00:00, January 1, 1970,
623 because that is OS- and implementation-specific. It is better to
624 store a date in an unambiguous representation. The ISO 8601 standard
625 defines YYYY-MM-DD as the date format, or YYYY-MM-DDTHH:MM:SS
626 (that's a literal "T" separating the date from the time).
627 Please do use the ISO 8601 instead of making us guess what
628 date 02/03/04 might be. ISO 8601 even sorts nicely as-is.
629 A text representation (like "1987-12-18") can be easily converted
630 into an OS-specific value using a module like C<Date::Parse>.
631 An array of values, such as those returned by C<localtime>, can be
632 converted to an OS-specific representation using C<Time::Local>.
634 When calculating specific times, such as for tests in time or date modules,
635 it may be appropriate to calculate an offset for the epoch.
638 my $offset = Time::Local::timegm(0, 0, 0, 1, 0, 70);
640 The value for C<$offset> in Unix will be C<0>, but in Mac OS Classic
641 will be some large number. C<$offset> can then be added to a Unix time
642 value to get what should be the proper value on any system.
644 =head2 Character sets and character encoding
646 Assume very little about character sets.
648 Assume nothing about numerical values (C<ord>, C<chr>) of characters.
649 Do not use explicit code point ranges (like C<\xHH-\xHH)>. However,
650 starting in Perl v5.22, regular expression pattern bracketed character
651 class ranges specified like C<qr/[\N{U+HH}-\N{U+HH}]/> are portable,
652 and starting in Perl v5.24, the same ranges are portable in C<tr///>.
653 You can portably use symbolic character classes like C<[:print:]>.
655 Do not assume that the alphabetic characters are encoded contiguously
656 (in the numeric sense). There may be gaps. Special coding in Perl,
657 however, guarantees that all subsets of C<qr/[A-Z]/>, C<qr/[a-z]/>, and
658 C<qr/[0-9]/> behave as expected. C<tr///> behaves the same for these
659 ranges. In patterns, any ranges specified with end points using the
660 C<\N{...}> notations ensures character set portability, but it is a bug
661 in Perl v5.22, that this isn't true of C<tr///>, fixed in v5.24.
663 Do not assume anything about the ordering of the characters.
664 The lowercase letters may come before or after the uppercase letters;
665 the lowercase and uppercase may be interlaced so that both "a" and "A"
666 come before "b"; the accented and other international characters may
667 be interlaced so that E<auml> comes before "b".
668 L<Unicode::Collate> can be used to sort this all out.
670 =head2 Internationalisation
672 If you may assume POSIX (a rather large assumption), you may read
673 more about the POSIX locale system from L<perllocale>. The locale
674 system at least attempts to make things a little bit more portable,
675 or at least more convenient and native-friendly for non-English
676 users. The system affects character sets and encoding, and date
677 and time formatting--amongst other things.
679 If you really want to be international, you should consider Unicode.
680 See L<perluniintro> and L<perlunicode> for more information.
682 If you want to use non-ASCII bytes (outside the bytes 0x00..0x7f) in
683 the "source code" of your code, to be portable you have to be explicit
684 about what bytes they are. Someone might for example be using your
685 code under a UTF-8 locale, in which case random native bytes might be
686 illegal ("Malformed UTF-8 ...") This means that for example embedding
687 ISO 8859-1 bytes beyond 0x7f into your strings might cause trouble
688 later. If the bytes are native 8-bit bytes, you can use the C<bytes>
689 pragma. If the bytes are in a string (regular expressions being
690 curious strings), you can often also use the C<\xHH> or more portably,
691 the C<\N{U+HH}> notations instead
692 of embedding the bytes as-is. If you want to write your code in UTF-8,
695 =head2 System Resources
697 If your code is destined for systems with severely constrained (or
698 missing!) virtual memory systems then you want to be I<especially> mindful
699 of avoiding wasteful constructs such as:
701 my @lines = <$very_large_file>; # bad
703 while (<$fh>) {$file .= $_} # sometimes bad
704 my $file = join('', <$fh>); # better
706 The last two constructs may appear unintuitive to most people. The
707 first repeatedly grows a string, whereas the second allocates a
708 large chunk of memory in one go. On some systems, the second is
709 more efficient than the first.
713 Most multi-user platforms provide basic levels of security, usually
714 implemented at the filesystem level. Some, however, unfortunately do
715 not. Thus the notion of user id, or "home" directory,
716 or even the state of being logged-in, may be unrecognizable on many
717 platforms. If you write programs that are security-conscious, it
718 is usually best to know what type of system you will be running
719 under so that you can write code explicitly for that platform (or
722 Don't assume the Unix filesystem access semantics: the operating
723 system or the filesystem may be using some ACL systems, which are
724 richer languages than the usual C<rwx>. Even if the C<rwx> exist,
725 their semantics might be different.
727 (From the security viewpoint, testing for permissions before attempting to
728 do something is silly anyway: if one tries this, there is potential
729 for race conditions. Someone or something might change the
730 permissions between the permissions check and the actual operation.
731 Just try the operation.)
733 Don't assume the Unix user and group semantics: especially, don't
734 expect C<< $< >> and C<< $> >> (or C<$(> and C<$)>) to work
735 for switching identities (or memberships).
737 Don't assume set-uid and set-gid semantics. (And even if you do,
738 think twice: set-uid and set-gid are a known can of security worms.)
742 For those times when it is necessary to have platform-specific code,
743 consider keeping the platform-specific code in one place, making porting
744 to other platforms easier. Use the C<Config> module and the special
745 variable C<$^O> to differentiate platforms, as described in
748 Be careful in the tests you supply with your module or programs.
749 Module code may be fully portable, but its tests might not be. This
750 often happens when tests spawn off other processes or call external
751 programs to aid in the testing, or when (as noted above) the tests
752 assume certain things about the filesystem and paths. Be careful not
753 to depend on a specific output style for errors, such as when checking
754 C<$!> after a failed system call. Using C<$!> for anything else than
755 displaying it as output is doubtful (though see the C<Errno> module for
756 testing reasonably portably for error value). Some platforms expect
757 a certain output format, and Perl on those platforms may have been
758 adjusted accordingly. Most specifically, don't anchor a regex when
759 testing an error value.
763 Modules uploaded to CPAN are tested by a variety of volunteers on
764 different platforms. These CPAN testers are notified by mail of each
765 new upload, and reply to the list with PASS, FAIL, NA (not applicable to
766 this platform), or UNKNOWN (unknown), along with any relevant notations.
768 The purpose of the testing is twofold: one, to help developers fix any
769 problems in their code that crop up because of lack of testing on other
770 platforms; two, to provide users with information about whether
771 a given module works on a given platform.
779 Mailing list: cpan-testers-discuss@perl.org
783 Testing results: L<http://www.cpantesters.org/>
789 Perl is built with a C<$^O> variable that indicates the operating
790 system it was built on. This was implemented
791 to help speed up code that would otherwise have to C<use Config>
792 and use the value of C<$Config{osname}>. Of course, to get more
793 detailed information about the system, looking into C<%Config> is
794 certainly recommended.
796 C<%Config> cannot always be trusted, however, because it was built
797 at compile time. If perl was built in one place, then transferred
798 elsewhere, some values may be wrong. The values may even have been
799 edited after the fact.
803 Perl works on a bewildering variety of Unix and Unix-like platforms (see
804 e.g. most of the files in the F<hints/> directory in the source code kit).
805 On most of these systems, the value of C<$^O> (hence C<$Config{'osname'}>,
806 too) is determined either by lowercasing and stripping punctuation from the
807 first field of the string returned by typing C<uname -a> (or a similar command)
808 at the shell prompt or by testing the file system for the presence of
809 uniquely named files such as a kernel or header file. Here, for example,
810 are a few of the more popular Unix flavors:
812 uname $^O $Config{'archname'}
813 --------------------------------------------
815 BSD/OS bsdos i386-bsdos
817 DYNIX/ptx dynixptx i386-dynixptx
818 FreeBSD freebsd freebsd-i386
819 Haiku haiku BePC-haiku
820 Linux linux arm-linux
821 Linux linux armv5tel-linux
822 Linux linux i386-linux
823 Linux linux i586-linux
824 Linux linux ppc-linux
825 HP-UX hpux PA-RISC1.1
827 Mac OS X darwin darwin
829 NeXT 4 next OPENSTEP-Mach
830 openbsd openbsd i386-openbsd
831 OSF1 dec_osf alpha-dec_osf
832 reliantunix-n svr4 RM400-svr4
833 SCO_SV sco_sv i386-sco_sv
834 SINIX-N svr4 RM400-svr4
835 sn4609 unicos CRAY_C90-unicos
836 sn6521 unicosmk t3e-unicosmk
837 sn9617 unicos CRAY_J90-unicos
838 SunOS solaris sun4-solaris
839 SunOS solaris i86pc-solaris
840 SunOS4 sunos sun4-sunos
842 Because the value of C<$Config{archname}> may depend on the
843 hardware architecture, it can vary more than the value of C<$^O>.
845 =head2 DOS and Derivatives
847 Perl has long been ported to Intel-style microcomputers running under
848 systems like PC-DOS, MS-DOS, OS/2, and most Windows platforms you can
849 bring yourself to mention (except for Windows CE, if you count that).
850 Users familiar with I<COMMAND.COM> or I<CMD.EXE> style shells should
851 be aware that each of these file specifications may have subtle
854 my $filespec0 = "c:/foo/bar/file.txt";
855 my $filespec1 = "c:\\foo\\bar\\file.txt";
856 my $filespec2 = 'c:\foo\bar\file.txt';
857 my $filespec3 = 'c:\\foo\\bar\\file.txt';
859 System calls accept either C</> or C<\> as the path separator.
860 However, many command-line utilities of DOS vintage treat C</> as
861 the option prefix, so may get confused by filenames containing C</>.
862 Aside from calling any external programs, C</> will work just fine,
863 and probably better, as it is more consistent with popular usage,
864 and avoids the problem of remembering what to backwhack and what
867 The DOS FAT filesystem can accommodate only "8.3" style filenames. Under
868 the "case-insensitive, but case-preserving" HPFS (OS/2) and NTFS (NT)
869 filesystems you may have to be careful about case returned with functions
870 like C<readdir> or used with functions like C<open> or C<opendir>.
872 DOS also treats several filenames as special, such as AUX, PRN,
873 NUL, CON, COM1, LPT1, LPT2, etc. Unfortunately, sometimes these
874 filenames won't even work if you include an explicit directory
875 prefix. It is best to avoid such filenames, if you want your code
876 to be portable to DOS and its derivatives. It's hard to know what
877 these all are, unfortunately.
879 Users of these operating systems may also wish to make use of
880 scripts such as I<pl2bat.bat> or I<pl2cmd> to
881 put wrappers around your scripts.
883 Newline (C<\n>) is translated as C<\015\012> by STDIO when reading from
884 and writing to files (see L<"Newlines">). C<binmode(FILEHANDLE)>
885 will keep C<\n> translated as C<\012> for that filehandle. Since it is a
886 no-op on other systems, C<binmode> should be used for cross-platform code
887 that deals with binary data. That's assuming you realize in advance
888 that your data is in binary. General-purpose programs should
889 often assume nothing about their data.
891 The C<$^O> variable and the C<$Config{archname}> values for various
892 DOSish perls are as follows:
894 OS $^O $Config{archname} ID Version
895 --------------------------------------------------------
899 Windows 3.1 ? ? 0 3 01
900 Windows 95 MSWin32 MSWin32-x86 1 4 00
901 Windows 98 MSWin32 MSWin32-x86 1 4 10
902 Windows ME MSWin32 MSWin32-x86 1 ?
903 Windows NT MSWin32 MSWin32-x86 2 4 xx
904 Windows NT MSWin32 MSWin32-ALPHA 2 4 xx
905 Windows NT MSWin32 MSWin32-ppc 2 4 xx
906 Windows 2000 MSWin32 MSWin32-x86 2 5 00
907 Windows XP MSWin32 MSWin32-x86 2 5 01
908 Windows 2003 MSWin32 MSWin32-x86 2 5 02
909 Windows Vista MSWin32 MSWin32-x86 2 6 00
910 Windows 7 MSWin32 MSWin32-x86 2 6 01
911 Windows 7 MSWin32 MSWin32-x64 2 6 01
912 Windows 2008 MSWin32 MSWin32-x86 2 6 01
913 Windows 2008 MSWin32 MSWin32-x64 2 6 01
914 Windows CE MSWin32 ? 3
917 The various MSWin32 Perl's can distinguish the OS they are running on
918 via the value of the fifth element of the list returned from
919 C<Win32::GetOSVersion()>. For example:
921 if ($^O eq 'MSWin32') {
922 my @os_version_info = Win32::GetOSVersion();
923 print +('3.1','95','NT')[$os_version_info[4]],"\n";
926 There are also C<Win32::IsWinNT()> and C<Win32::IsWin95()>; try C<perldoc Win32>,
927 and as of libwin32 0.19 (not part of the core Perl distribution)
928 C<Win32::GetOSName()>. The very portable C<POSIX::uname()> will work too:
930 c:\> perl -MPOSIX -we "print join '|', uname"
931 Windows NT|moonru|5.0|Build 2195 (Service Pack 2)|x86
939 The djgpp environment for DOS, L<http://www.delorie.com/djgpp/>
944 The EMX environment for DOS, OS/2, etc. emx@iaehv.nl,
945 L<ftp://hobbes.nmsu.edu/pub/os2/dev/emx/> Also L<perlos2>.
949 Build instructions for Win32 in L<perlwin32>, or under the Cygnus environment
954 The C<Win32::*> modules in L<Win32>.
958 The ActiveState Pages, L<http://www.activestate.com/>
962 The Cygwin environment for Win32; F<README.cygwin> (installed
963 as L<perlcygwin>), L<http://www.cygwin.com/>
967 The U/WIN environment for Win32,
968 L<http://www.research.att.com/sw/tools/uwin/>
972 Build instructions for OS/2, L<perlos2>
978 Perl on VMS is discussed in L<perlvms> in the Perl distribution.
980 The official name of VMS as of this writing is OpenVMS.
982 Interacting with Perl from the Digital Command Language (DCL) shell
983 often requires a different set of quotation marks than Unix shells do.
986 $ perl -e "print ""Hello, world.\n"""
989 There are several ways to wrap your Perl scripts in DCL F<.COM> files, if
990 you are so inclined. For example:
992 $ write sys$output "Hello from DCL!"
994 $ then perl -x 'f$environment("PROCEDURE")
995 $ else perl -x - 'p1 'p2 'p3 'p4 'p5 'p6 'p7 'p8
996 $ deck/dollars="__END__"
999 print "Hello from Perl!\n";
1004 Do take care with C<$ ASSIGN/nolog/user SYS$COMMAND: SYS$INPUT> if your
1005 Perl-in-DCL script expects to do things like C<< $read = <STDIN>; >>.
1007 The VMS operating system has two filesystems, designated by their
1008 on-disk structure (ODS) level: ODS-2 and its successor ODS-5. The
1009 initial port of Perl to VMS pre-dates ODS-5, but all current testing and
1010 development assumes ODS-5 and its capabilities, including case
1011 preservation, extended characters in filespecs, and names up to 8192
1014 Perl on VMS can accept either VMS- or Unix-style file
1015 specifications as in either of the following:
1017 $ perl -ne "print if /perl_setup/i" SYS$LOGIN:LOGIN.COM
1018 $ perl -ne "print if /perl_setup/i" /sys$login/login.com
1020 but not a mixture of both as in:
1022 $ perl -ne "print if /perl_setup/i" sys$login:/login.com
1023 Can't open sys$login:/login.com: file specification syntax error
1025 In general, the easiest path to portability is always to specify
1026 filenames in Unix format unless they will need to be processed by native
1027 commands or utilities. Because of this latter consideration, the
1028 File::Spec module by default returns native format specifications
1029 regardless of input format. This default may be reversed so that
1030 filenames are always reported in Unix format by specifying the
1031 C<DECC$FILENAME_UNIX_REPORT> feature logical in the environment.
1033 The file type, or extension, is always present in a VMS-format file
1034 specification even if it's zero-length. This means that, by default,
1035 C<readdir> will return a trailing dot on a file with no extension, so
1036 where you would see C<"a"> on Unix you'll see C<"a."> on VMS. However,
1037 the trailing dot may be suppressed by enabling the
1038 C<DECC$READDIR_DROPDOTNOTYPE> feature in the environment (see the CRTL
1039 documentation on feature logical names).
1041 What C<\n> represents depends on the type of file opened. It usually
1042 represents C<\012> but it could also be C<\015>, C<\012>, C<\015\012>,
1043 C<\000>, C<\040>, or nothing depending on the file organization and
1044 record format. The C<VMS::Stdio> module provides access to the
1045 special C<fopen()> requirements of files with unusual attributes on VMS.
1047 The value of C<$^O> on OpenVMS is "VMS". To determine the architecture
1048 that you are running on refer to C<$Config{'archname'}>.
1050 On VMS, perl determines the UTC offset from the C<SYS$TIMEZONE_DIFFERENTIAL>
1051 logical name. Although the VMS epoch began at 17-NOV-1858 00:00:00.00,
1052 calls to C<localtime> are adjusted to count offsets from
1053 01-JAN-1970 00:00:00.00, just like Unix.
1061 F<README.vms> (installed as F<README_vms>), L<perlvms>
1065 vmsperl list, vmsperl-subscribe@perl.org
1069 vmsperl on the web, L<http://www.sidhe.org/vmsperl/index.html>
1073 VMS Software Inc. web site, L<http://www.vmssoftware.com>
1079 Perl on VOS (also known as OpenVOS) is discussed in F<README.vos>
1080 in the Perl distribution (installed as L<perlvos>). Perl on VOS
1081 can accept either VOS- or Unix-style file specifications as in
1082 either of the following:
1084 $ perl -ne "print if /perl_setup/i" >system>notices
1085 $ perl -ne "print if /perl_setup/i" /system/notices
1087 or even a mixture of both as in:
1089 $ perl -ne "print if /perl_setup/i" >system/notices
1091 Even though VOS allows the slash character to appear in object
1092 names, because the VOS port of Perl interprets it as a pathname
1093 delimiting character, VOS files, directories, or links whose
1094 names contain a slash character cannot be processed. Such files
1095 must be renamed before they can be processed by Perl.
1097 Older releases of VOS (prior to OpenVOS Release 17.0) limit file
1098 names to 32 or fewer characters, prohibit file names from
1099 starting with a C<-> character, and prohibit file names from
1100 containing any character matching C<< tr/ !#%&'()*;<=>?// >>.
1102 Newer releases of VOS (OpenVOS Release 17.0 or later) support a
1103 feature known as extended names. On these releases, file names
1104 can contain up to 255 characters, are prohibited from starting
1105 with a C<-> character, and the set of prohibited characters is
1106 reduced to any character matching C<< tr/#%*<>?// >>. There are
1107 restrictions involving spaces and apostrophes: these characters
1108 must not begin or end a name, nor can they immediately precede or
1109 follow a period. Additionally, a space must not immediately
1110 precede another space or hyphen. Specifically, the following
1111 character combinations are prohibited: space-space,
1112 space-hyphen, period-space, space-period, period-apostrophe,
1113 apostrophe-period, leading or trailing space, and leading or
1114 trailing apostrophe. Although an extended file name is limited
1115 to 255 characters, a path name is still limited to 256
1118 The value of C<$^O> on VOS is "vos". To determine the
1119 architecture that you are running on without resorting to loading
1120 all of C<%Config> you can examine the content of the C<@INC> array
1124 print "I'm on a Stratus box!\n";
1126 print "I'm not on a Stratus box!\n";
1136 F<README.vos> (installed as L<perlvos>)
1140 The VOS mailing list.
1142 There is no specific mailing list for Perl on VOS. You can contact
1143 the Stratus Technologies Customer Assistance Center (CAC) for your
1144 region, or you can use the contact information located in the
1145 distribution files on the Stratus Anonymous FTP site.
1149 Stratus Technologies on the web at L<http://www.stratus.com>
1153 VOS Open-Source Software on the web at L<http://ftp.stratus.com/pub/vos/vos.html>
1157 =head2 EBCDIC Platforms
1159 v5.22 core Perl runs on z/OS (formerly OS/390). Theoretically it could
1160 run on the successors of OS/400 on AS/400 minicomputers as well as
1161 VM/ESA, and BS2000 for S/390 Mainframes. Such computers use EBCDIC
1162 character sets internally (usually
1163 Character Code Set ID 0037 for OS/400 and either 1047 or POSIX-BC for S/390
1166 The rest of this section may need updating, but we don't know what it
1167 should say. Please email comments to
1168 L<perlbug@perl.org|mailto:perlbug@perl.org>.
1170 On the mainframe Perl currently works under the "Unix system
1171 services for OS/390" (formerly known as OpenEdition), VM/ESA OpenEdition, or
1172 the BS200 POSIX-BC system (BS2000 is supported in Perl 5.6 and greater).
1173 See L<perlos390> for details. Note that for OS/400 there is also a port of
1174 Perl 5.8.1/5.10.0 or later to the PASE which is ASCII-based (as opposed to
1175 ILE which is EBCDIC-based), see L<perlos400>.
1177 As of R2.5 of USS for OS/390 and Version 2.3 of VM/ESA these Unix
1178 sub-systems do not support the C<#!> shebang trick for script invocation.
1179 Hence, on OS/390 and VM/ESA Perl scripts can be executed with a header
1180 similar to the following simple script:
1183 eval 'exec /usr/local/bin/perl -S $0 ${1+"$@"}'
1185 #!/usr/local/bin/perl # just a comment really
1187 print "Hello from perl!\n";
1189 OS/390 will support the C<#!> shebang trick in release 2.8 and beyond.
1190 Calls to C<system> and backticks can use POSIX shell syntax on all
1193 On the AS/400, if PERL5 is in your library list, you may need
1194 to wrap your Perl scripts in a CL procedure to invoke them like so:
1197 CALL PGM(PERL5/PERL) PARM('/QOpenSys/hello.pl')
1200 This will invoke the Perl script F<hello.pl> in the root of the
1201 QOpenSys file system. On the AS/400 calls to C<system> or backticks
1204 On these platforms, bear in mind that the EBCDIC character set may have
1205 an effect on what happens with some Perl functions (such as C<chr>,
1206 C<pack>, C<print>, C<printf>, C<ord>, C<sort>, C<sprintf>, C<unpack>), as
1207 well as bit-fiddling with ASCII constants using operators like C<^>, C<&>
1208 and C<|>, not to mention dealing with socket interfaces to ASCII computers
1209 (see L<"Newlines">).
1211 Fortunately, most web servers for the mainframe will correctly
1212 translate the C<\n> in the following statement to its ASCII equivalent
1213 (C<\r> is the same under both Unix and z/OS):
1215 print "Content-type: text/html\r\n\r\n";
1217 The values of C<$^O> on some of these platforms includes:
1219 uname $^O $Config{'archname'}
1220 --------------------------------------------
1223 POSIX-BC posix-bc BS2000-posix-bc
1225 Some simple tricks for determining if you are running on an EBCDIC
1226 platform could include any of the following (perhaps all):
1228 if ("\t" eq "\005") { print "EBCDIC may be spoken here!\n"; }
1230 if (ord('A') == 193) { print "EBCDIC may be spoken here!\n"; }
1232 if (chr(169) eq 'z') { print "EBCDIC may be spoken here!\n"; }
1234 One thing you may not want to rely on is the EBCDIC encoding
1235 of punctuation characters since these may differ from code page to code
1236 page (and once your module or script is rumoured to work with EBCDIC,
1237 folks will want it to work with all EBCDIC character sets).
1245 L<perlos390>, L<perlos400>, L<perlbs2000>, L<perlebcdic>.
1249 The perl-mvs@perl.org list is for discussion of porting issues as well as
1250 general usage issues for all EBCDIC Perls. Send a message body of
1251 "subscribe perl-mvs" to majordomo@perl.org.
1255 AS/400 Perl information at
1256 L<http://as400.rochester.ibm.com/>
1257 as well as on CPAN in the F<ports/> directory.
1261 =head2 Acorn RISC OS
1263 Because Acorns use ASCII with newlines (C<\n>) in text files as C<\012> like
1264 Unix, and because Unix filename emulation is turned on by default,
1265 most simple scripts will probably work "out of the box". The native
1266 filesystem is modular, and individual filesystems are free to be
1267 case-sensitive or insensitive, and are usually case-preserving. Some
1268 native filesystems have name length limits, which file and directory
1269 names are silently truncated to fit. Scripts should be aware that the
1270 standard filesystem currently has a name length limit of B<10>
1271 characters, with up to 77 items in a directory, but other filesystems
1272 may not impose such limitations.
1274 Native filenames are of the form
1276 Filesystem#Special_Field::DiskName.$.Directory.Directory.File
1280 Special_Field is not usually present, but may contain . and $ .
1281 Filesystem =~ m|[A-Za-z0-9_]|
1282 DsicName =~ m|[A-Za-z0-9_/]|
1283 $ represents the root directory
1284 . is the path separator
1285 @ is the current directory (per filesystem but machine global)
1286 ^ is the parent directory
1287 Directory and File =~ m|[^\0- "\.\$\%\&:\@\\^\|\177]+|
1289 The default filename translation is roughly C<tr|/.|./|;>
1291 Note that C<"ADFS::HardDisk.$.File" ne 'ADFS::HardDisk.$.File'> and that
1292 the second stage of C<$> interpolation in regular expressions will fall
1293 foul of the C<$.> if scripts are not careful.
1295 Logical paths specified by system variables containing comma-separated
1296 search lists are also allowed; hence C<System:Modules> is a valid
1297 filename, and the filesystem will prefix C<Modules> with each section of
1298 C<System$Path> until a name is made that points to an object on disk.
1299 Writing to a new file C<System:Modules> would be allowed only if
1300 C<System$Path> contains a single item list. The filesystem will also
1301 expand system variables in filenames if enclosed in angle brackets, so
1302 C<< <System$Dir>.Modules >> would look for the file
1303 S<C<$ENV{'System$Dir'} . 'Modules'>>. The obvious implication of this is
1304 that B<fully qualified filenames can start with C<< <> >>> and should
1305 be protected when C<open> is used for input.
1307 Because C<.> was in use as a directory separator and filenames could not
1308 be assumed to be unique after 10 characters, Acorn implemented the C
1309 compiler to strip the trailing C<.c> C<.h> C<.s> and C<.o> suffix from
1310 filenames specified in source code and store the respective files in
1311 subdirectories named after the suffix. Hence files are translated:
1314 C:foo.h C:h.foo (logical path variable)
1315 sys/os.h sys.h.os (C compiler groks Unix-speak)
1316 10charname.c c.10charname
1317 10charname.o o.10charname
1318 11charname_.c c.11charname (assuming filesystem truncates at 10)
1320 The Unix emulation library's translation of filenames to native assumes
1321 that this sort of translation is required, and it allows a user-defined list
1322 of known suffixes that it will transpose in this fashion. This may
1323 seem transparent, but consider that with these rules F<foo/bar/baz.h>
1324 and F<foo/bar/h/baz> both map to F<foo.bar.h.baz>, and that C<readdir> and
1325 C<glob> cannot and do not attempt to emulate the reverse mapping. Other
1326 C<.>'s in filenames are translated to C</>.
1328 As implied above, the environment accessed through C<%ENV> is global, and
1329 the convention is that program specific environment variables are of the
1330 form C<Program$Name>. Each filesystem maintains a current directory,
1331 and the current filesystem's current directory is the B<global> current
1332 directory. Consequently, sociable programs don't change the current
1333 directory but rely on full pathnames, and programs (and Makefiles) cannot
1334 assume that they can spawn a child process which can change the current
1335 directory without affecting its parent (and everyone else for that
1338 Because native operating system filehandles are global and are currently
1339 allocated down from 255, with 0 being a reserved value, the Unix emulation
1340 library emulates Unix filehandles. Consequently, you can't rely on
1341 passing C<STDIN>, C<STDOUT>, or C<STDERR> to your children.
1343 The desire of users to express filenames of the form
1344 C<< <Foo$Dir>.Bar >> on the command line unquoted causes problems,
1345 too: C<``> command output capture has to perform a guessing game. It
1346 assumes that a string C<< <[^<>]+\$[^<>]> >> is a
1347 reference to an environment variable, whereas anything else involving
1348 C<< < >> or C<< > >> is redirection, and generally manages to be 99%
1349 right. Of course, the problem remains that scripts cannot rely on any
1350 Unix tools being available, or that any tools found have Unix-like command
1353 Extensions and XS are, in theory, buildable by anyone using free
1354 tools. In practice, many don't, as users of the Acorn platform are
1355 used to binary distributions. MakeMaker does run, but no available
1356 make currently copes with MakeMaker's makefiles; even if and when
1357 this should be fixed, the lack of a Unix-like shell will cause
1358 problems with makefile rules, especially lines of the form C<cd
1359 sdbm && make all>, and anything using quoting.
1361 "S<RISC OS>" is the proper name for the operating system, but the value
1362 in C<$^O> is "riscos" (because we don't like shouting).
1366 Perl has been ported to many platforms that do not fit into any of
1367 the categories listed above. Some, such as AmigaOS,
1368 QNX, Plan 9, and VOS, have been well-integrated into the standard
1369 Perl source code kit. You may need to see the F<ports/> directory
1370 on CPAN for information, and possibly binaries, for the likes of:
1371 aos, Atari ST, lynxos, riscos, Novell Netware, Tandem Guardian,
1372 I<etc.> (Yes, we know that some of these OSes may fall under the
1373 Unix category, but we are not a standards body.)
1375 Some approximate operating system names and their C<$^O> values
1376 in the "OTHER" category include:
1378 OS $^O $Config{'archname'}
1379 ------------------------------------------
1380 Amiga DOS amigaos m68k-amigos
1388 Amiga, F<README.amiga> (installed as L<perlamiga>).
1392 A free perl5-based PERL.NLM for Novell Netware is available in
1393 precompiled binary and source code form from L<http://www.novell.com/>
1394 as well as from CPAN.
1398 S<Plan 9>, F<README.plan9>
1402 =head1 FUNCTION IMPLEMENTATIONS
1404 Listed below are functions that are either completely unimplemented
1405 or else have been implemented differently on various platforms.
1406 Following each description will be, in parentheses, a list of
1407 platforms that the description applies to.
1409 The list may well be incomplete, or even wrong in some places. When
1410 in doubt, consult the platform-specific README files in the Perl
1411 source distribution, and any other documentation resources accompanying
1414 Be aware, moreover, that even among Unix-ish systems there are variations.
1416 For many functions, you can also query C<%Config>, exported by
1417 default from the C<Config> module. For example, to check whether the
1418 platform has the C<lstat> call, check C<$Config{d_lstat}>. See
1419 L<Config> for a full description of available variables.
1421 =head2 Alphabetical Listing of Perl Functions
1427 C<-w> only inspects the read-only file attribute (FILE_ATTRIBUTE_READONLY),
1428 which determines whether the directory can be deleted, not whether it can
1429 be written to. Directories always have read and write access unless denied
1430 by discretionary access control lists (DACLs). (S<Win32>)
1432 C<-r>, C<-w>, C<-x>, and C<-o> tell whether the file is accessible,
1433 which may not reflect UIC-based file protections. (VMS)
1435 C<-s> by name on an open file will return the space reserved on disk,
1436 rather than the current extent. C<-s> on an open filehandle returns the
1437 current size. (S<RISC OS>)
1439 C<-R>, C<-W>, C<-X>, C<-O> are indistinguishable from C<-r>, C<-w>,
1440 C<-x>, C<-o>. (Win32, VMS, S<RISC OS>)
1442 C<-g>, C<-k>, C<-l>, C<-u>, C<-A> are not particularly meaningful.
1443 (Win32, VMS, S<RISC OS>)
1445 C<-p> is not particularly meaningful. (VMS, S<RISC OS>)
1447 C<-d> is true if passed a device spec without an explicit directory.
1450 C<-x> (or C<-X>) determine if a file ends in one of the executable
1451 suffixes. C<-S> is meaningless. (Win32)
1453 C<-x> (or C<-X>) determine if a file has an executable file type.
1458 Emulated using timers that must be explicitly polled whenever Perl
1459 wants to dispatch "safe signals" and therefore cannot interrupt
1460 blocking system calls. (Win32)
1464 Due to issues with various CPUs, math libraries, compilers, and standards,
1465 results for C<atan2()> may vary depending on any combination of the above.
1466 Perl attempts to conform to the Open Group/IEEE standards for the results
1467 returned from C<atan2()>, but cannot force the issue if the system Perl is
1468 run on does not allow it. (Tru64, HP-UX 10.20)
1470 The current version of the standards for C<atan2()> is available at
1471 L<http://www.opengroup.org/onlinepubs/009695399/functions/atan2.html>.
1475 Meaningless. (S<RISC OS>)
1477 Reopens file and restores pointer; if function fails, underlying
1478 filehandle may be closed, or pointer may be in a different position.
1481 The value returned by C<tell> may be affected after the call, and
1482 the filehandle may be flushed. (Win32)
1486 Only good for changing "owner" read-write access, "group", and "other"
1487 bits are meaningless. (Win32)
1489 Only good for changing "owner" and "other" read-write access. (S<RISC OS>)
1491 Access permissions are mapped onto VOS access-control list changes. (VOS)
1493 The actual permissions set depend on the value of the C<CYGWIN>
1494 in the SYSTEM environment settings. (Cygwin)
1496 Setting the exec bit on some locations (generally F</sdcard>) will return true
1497 but not actually set the bit. (Android)
1501 Not implemented. (Win32, S<Plan 9>, S<RISC OS>)
1503 Does nothing, but won't fail. (Win32)
1505 A little funky, because VOS's notion of ownership is a little funky (VOS).
1509 Not implemented. (Win32, VMS, S<Plan 9>, S<RISC OS>, VOS)
1513 May not be available if library or source was not provided when building
1516 Not implemented. (Android)
1520 Not implemented. (VMS, S<Plan 9>, VOS)
1524 Not implemented. (VMS, S<Plan 9>, VOS)
1528 Not useful. (S<RISC OS>)
1530 Not supported. (Cygwin, Win32)
1532 Invokes VMS debugger. (VMS)
1536 C<exec LIST> without the use of indirect object syntax (C<exec PROGRAM LIST>)
1537 may fall back to trying the shell if the first C<spawn()> fails. (Win32)
1539 Does not automatically flush output handles on some platforms.
1540 (SunOS, Solaris, HP-UX)
1542 Not supported. (Symbian OS)
1546 Emulates Unix C<exit()> (which considers C<exit 1> to indicate an error) by
1547 mapping the C<1> to C<SS$_ABORT> (C<44>). This behavior may be overridden
1548 with the pragma C<use vmsish 'exit'>. As with the CRTL's C<exit()>
1549 function, C<exit 0> is also mapped to an exit status of C<SS$_NORMAL>
1550 (C<1>); this mapping cannot be overridden. Any other argument to
1552 is used directly as Perl's exit status. On VMS, unless the future
1553 POSIX_EXIT mode is enabled, the exit code should always be a valid
1554 VMS exit code and not a generic number. When the POSIX_EXIT mode is
1555 enabled, a generic number will be encoded in a method compatible with
1556 the C library _POSIX_EXIT macro so that it can be decoded by other
1557 programs, particularly ones written in C, like the GNV package. (VMS)
1559 C<exit()> resets file pointers, which is a problem when called
1560 from a child process (created by C<fork()>) in C<BEGIN>.
1561 A workaround is to use C<POSIX::_exit>. (Solaris)
1563 exit unless $Config{archname} =~ /\bsolaris\b/;
1564 require POSIX and POSIX::_exit(0);
1568 Not implemented. (Win32)
1570 Some functions available based on the version of VMS. (VMS)
1574 Not implemented (VMS, S<RISC OS>, VOS).
1578 Not implemented. (AmigaOS, S<RISC OS>, VMS)
1580 Emulated using multiple interpreters. See L<perlfork>. (Win32)
1582 Does not automatically flush output handles on some platforms.
1583 (SunOS, Solaris, HP-UX)
1587 Not implemented. (S<RISC OS>)
1591 Not implemented. (Win32, VMS, S<RISC OS>)
1595 Not implemented. (Win32, S<RISC OS>)
1599 Not implemented. (Win32, VMS, S<RISC OS>, VOS)
1603 Not implemented. (Win32)
1605 Not useful. (S<RISC OS>)
1609 Not implemented. (Win32, VMS, S<RISC OS>)
1613 Not implemented. (Android, Win32, S<Plan 9>)
1617 Not implemented. (Win32)
1619 Not useful. (S<RISC OS>)
1623 Not implemented. (Win32, VMS, S<RISC OS>)
1627 Not implemented. (Android, Win32, S<Plan 9>)
1629 =item getprotobynumber
1631 Not implemented. (Android)
1637 Not implemented. (Android, Win32)
1641 Not implemented. (Android, Win32, VMS)
1645 C<gethostbyname('localhost')> does not work everywhere: you may have
1646 to use C<gethostbyname('127.0.0.1')>. (S<Irix 5>)
1650 Not implemented. (Win32)
1654 Not implemented. (Android, Win32, S<Plan 9>)
1658 Not implemented. (Android, Win32, S<Plan 9>)
1662 Not implemented. (Win32, S<Plan 9>)
1666 Not implemented. (Android)
1670 Not implemented. (Android, Win32, S<Plan 9>, S<RISC OS>)
1674 Not implemented. (Win32, S<Plan 9>, S<RISC OS>)
1678 Not implemented. (Android, Win32, S<Plan 9>, S<RISC OS>)
1682 Not implemented. (S<Plan 9>, Win32, S<RISC OS>)
1686 Not implemented. (Win32)
1688 Either not implemented or a no-op. (Android)
1692 Not implemented. (Android, S<RISC OS>, VMS, Win32)
1696 Not implemented. (Android, Win32)
1700 Not implemented. (Android, Win32, S<Plan 9>)
1704 Not implemented. (Android, Win32, S<Plan 9>)
1708 Not implemented. (S<Plan 9>, Win32)
1710 =item getsockopt SOCKET,LEVEL,OPTNAME
1712 Not implemented. (S<Plan 9>)
1716 This operator is implemented via the C<File::Glob> extension on most
1717 platforms. See L<File::Glob> for portability information.
1721 In theory, C<gmtime()> is reliable from -2**63 to 2**63-1. However,
1722 because work arounds in the implementation use floating point numbers,
1723 it will become inaccurate as the time gets larger. This is a bug and
1724 will be fixed in the future.
1726 On VOS, time values are 32-bit quantities.
1728 =item ioctl FILEHANDLE,FUNCTION,SCALAR
1730 Not implemented. (VMS)
1732 Available only for socket handles, and it does what the C<ioctlsocket()> call
1733 in the Winsock API does. (Win32)
1735 Available only for socket handles. (S<RISC OS>)
1739 Not implemented, hence not useful for taint checking. (S<RISC OS>)
1741 C<kill()> doesn't have the semantics of C<raise()>, i.e. it doesn't send
1742 a signal to the identified process like it does on Unix platforms.
1743 Instead C<kill($sig, $pid)> terminates the process identified by C<$pid>,
1744 and makes it exit immediately with exit status $sig. As in Unix, if
1745 $sig is 0 and the specified process exists, it returns true without
1746 actually terminating it. (Win32)
1748 C<kill(-9, $pid)> will terminate the process specified by C<$pid> and
1749 recursively all child processes owned by it. This is different from
1750 the Unix semantics, where the signal will be delivered to all
1751 processes in the same process group as the process specified by
1754 A pid of -1 indicating all processes on the system is not currently
1759 Not implemented. (S<RISC OS>, VOS)
1761 Link count not updated because hard links are not quite that hard
1762 (They are sort of half-way between hard and soft links). (AmigaOS)
1764 Hard links are implemented on Win32 under NTFS only. They are
1765 natively supported on Windows 2000 and later. On Windows NT they
1766 are implemented using the Windows POSIX subsystem support and the
1767 Perl process will need Administrator or Backup Operator privileges
1768 to create hard links.
1770 Available on 64 bit OpenVMS 8.2 and later. (VMS)
1774 localtime() has the same range as L</gmtime>, but because time zone
1775 rules change its accuracy for historical and future times may degrade
1776 but usually by no more than an hour.
1780 Not implemented. (S<RISC OS>)
1782 Return values (especially for device and inode) may be bogus. (Win32)
1792 Not implemented. (Android, Win32, VMS, S<Plan 9>, S<RISC OS>, VOS)
1796 open to C<|-> and C<-|> are unsupported. (Win32, S<RISC OS>)
1798 Opening a process does not automatically flush output handles on some
1799 platforms. (SunOS, Solaris, HP-UX)
1803 Not implemented. (Win32, VMS, S<RISC OS>)
1807 Can't move directories between directories on different logical volumes. (Win32)
1811 Will not cause C<readdir()> to re-read the directory stream. The entries
1812 already read before the C<rewinddir()> call will just be returned again
1813 from a cache buffer. (Win32)
1817 Only implemented on sockets. (Win32, VMS)
1819 Only reliable on sockets. (S<RISC OS>)
1821 Note that the C<select FILEHANDLE> form is generally portable.
1829 Not implemented. (Android, Win32, VMS, S<RISC OS>)
1833 Not implemented. (Android, VMS, Win32, S<RISC OS>)
1837 Not implemented. (Win32, VMS, S<RISC OS>, VOS)
1841 Not implemented. (Win32, VMS, S<RISC OS>, VOS)
1845 Not implemented. (Android, Win32, S<RISC OS>)
1849 Not implemented. (S<Plan 9>)
1859 Not implemented. (Android, Win32, VMS, S<RISC OS>)
1863 Emulated using synchronization functions such that it can be
1864 interrupted by C<alarm()>, and limited to a maximum of 4294967 seconds,
1865 approximately 49 days. (Win32)
1869 A relatively recent addition to socket functions, may not
1870 be implemented even in Unix platforms.
1874 Not implemented. (S<RISC OS>)
1876 Available on 64 bit OpenVMS 8.2 and later. (VMS)
1880 Platforms that do not have rdev, blksize, or blocks will return these
1881 as '', so numeric comparison or manipulation of these fields may cause
1882 'not numeric' warnings.
1884 ctime not supported on UFS (S<Mac OS X>).
1886 ctime is creation time instead of inode change time (Win32).
1888 device and inode are not meaningful. (Win32)
1890 device and inode are not necessarily reliable. (VMS)
1892 mtime, atime and ctime all return the last modification time. Device and
1893 inode are not necessarily reliable. (S<RISC OS>)
1895 dev, rdev, blksize, and blocks are not available. inode is not
1896 meaningful and will differ between stat calls on the same file. (os2)
1898 some versions of cygwin when doing a C<stat("foo")> and if not finding it
1899 may then attempt to C<stat("foo.exe")> (Cygwin)
1901 On Win32 C<stat()> needs to open the file to determine the link count
1902 and update attributes that may have been changed through hard links.
1903 Setting C<${^WIN32_SLOPPY_STAT}> to a true value speeds up C<stat()> by
1904 not performing this operation. (Win32)
1908 Not implemented. (Win32, S<RISC OS>)
1910 Implemented on 64 bit VMS 8.3. VMS requires the symbolic link to be in Unix
1911 syntax if it is intended to resolve to a valid path.
1915 Not implemented. (Win32, VMS, S<RISC OS>, VOS)
1919 The traditional "0", "1", and "2" MODEs are implemented with different
1920 numeric values on some systems. The flags exported by C<Fcntl>
1921 (O_RDONLY, O_WRONLY, O_RDWR) should work everywhere though. (S<Mac
1926 As an optimization, may not call the command shell specified in
1927 C<$ENV{PERL5SHELL}>. C<system(1, @args)> spawns an external
1928 process and immediately returns its process designator, without
1929 waiting for it to terminate. Return value may be used subsequently
1930 in C<wait> or C<waitpid>. Failure to C<spawn()> a subprocess is indicated
1931 by setting C<$?> to S<C<"255 << 8">>. C<$?> is set in a way compatible with
1932 Unix (i.e. the exitstatus of the subprocess is obtained by S<C<"$? >> 8">>,
1933 as described in the documentation). (Win32)
1935 There is no shell to process metacharacters, and the native standard is
1936 to pass a command line terminated by "\n" "\r" or "\0" to the spawned
1937 program. Redirection such as C<< > foo >> is performed (if at all) by
1938 the run time library of the spawned program. C<system> I<list> will call
1939 the Unix emulation library's C<exec> emulation, which attempts to provide
1940 emulation of the stdin, stdout, stderr in force in the parent, providing
1941 the child program uses a compatible version of the emulation library.
1942 I<scalar> will call the native command line direct and no such emulation
1943 of a child Unix program will exists. Mileage B<will> vary. (S<RISC OS>)
1945 C<system LIST> without the use of indirect object syntax (C<system PROGRAM LIST>)
1946 may fall back to trying the shell if the first C<spawn()> fails. (Win32)
1948 Does not automatically flush output handles on some platforms.
1949 (SunOS, Solaris, HP-UX)
1951 The return value is POSIX-like (shifted up by 8 bits), which only allows
1952 room for a made-up value derived from the severity bits of the native
1953 32-bit condition code (unless overridden by C<use vmsish 'status'>).
1954 If the native condition code is one that has a POSIX value encoded, the
1955 POSIX value will be decoded to extract the expected exit value.
1956 For more details see L<perlvms/$?>. (VMS)
1960 Not implemented. (Android)
1964 "cumulative" times will be bogus. On anything other than Windows NT
1965 or Windows 2000, "system" time will be bogus, and "user" time is
1966 actually the time returned by the C<clock()> function in the C runtime
1969 Not useful. (S<RISC OS>)
1973 Not implemented. (Older versions of VMS)
1975 Truncation to same-or-shorter lengths only. (VOS)
1977 If a FILEHANDLE is supplied, it must be writable and opened in append
1978 mode (i.e., use C<<< open(FH, '>>filename') >>>
1979 or C<sysopen(FH,...,O_APPEND|O_RDWR)>. If a filename is supplied, it
1980 should not be held open elsewhere. (Win32)
1984 Returns undef where unavailable.
1986 C<umask> works but the correct permissions are set only when the file
1987 is finally closed. (AmigaOS)
1991 Only the modification time is updated. (VMS, S<RISC OS>)
1993 May not behave as expected. Behavior depends on the C runtime
1994 library's implementation of C<utime()>, and the filesystem being
1995 used. The FAT filesystem typically does not support an "access
1996 time" field, and it may limit timestamps to a granularity of
1997 two seconds. (Win32)
2003 Can only be applied to process handles returned for processes spawned
2004 using C<system(1, ...)> or pseudo processes created with C<fork()>. (Win32)
2006 Not useful. (S<RISC OS>)
2011 =head1 Supported Platforms
2013 The following platforms are known to build Perl 5.12 (as of April 2010,
2014 its release date) from the standard source code distribution available
2015 at L<http://www.cpan.org/src>
2019 =item Linux (x86, ARM, IA64)
2033 =item Windows Server 2003
2037 =item Windows Server 2008
2045 Some tests are known to fail:
2051 F<ext/XS-APItes/t/call_checker.t> - see
2052 L<https://rt.perl.org/Ticket/Display.html?id=78502>
2056 F<dist/I18N-Collate/t/I18N-Collate.t>
2060 F<ext/Win32CORE/t/win32core.t> - may fail on recent cygwin installs.
2064 =item Solaris (x86, SPARC)
2070 =item Alpha (7.2 and later)
2072 =item I64 (8.2 and later)
2082 =item Debian GNU/kFreeBSD
2086 =item Irix (6.5. What else?)
2094 =item QNX Neutrino RTOS (6.5.0)
2098 =item Stratus OpenVOS (17.0 or later)
2104 =item time_t issues that may or may not be fixed
2108 =item Symbian (Series 60 v3, 3.2 and 5 - what else?)
2110 =item Stratus VOS / OpenVOS
2118 Perl now builds with FreeMiNT/Atari. It fails a few tests, that needs
2121 The FreeMiNT port uses GNU dld for loadable module capabilities. So
2122 ensure you have that library installed when building perl.
2126 =head1 EOL Platforms
2130 The following platforms were supported by a previous version of
2131 Perl but have been officially removed from Perl's source code
2142 The following platforms were supported up to 5.10. They may still
2143 have worked in 5.12, but supporting code has been removed for 5.14:
2159 The following platforms were supported by a previous version of
2160 Perl but have been officially removed from Perl's source code
2167 =item Apollo Domain/OS
2169 =item Apple Mac OS 8/9
2176 =head1 Supported Platforms (Perl 5.8)
2178 As of July 2002 (the Perl release 5.8.0), the following platforms were
2179 able to build Perl from the standard source code distribution
2180 available at L<http://www.cpan.org/src/>
2191 HI-UXMPP (Hitachi) (5.8.0 worked but we didn't know it)
2201 ReliantUNIX (formerly SINIX)
2203 OpenVMS (formerly VMS)
2204 Open UNIX (Unixware) (since Perl 5.8.1/5.9.0)
2206 OS/400 (using the PASE) (since Perl 5.8.1/5.9.0)
2208 POSIX-BC (formerly BS2000)
2213 Tru64 UNIX (formerly DEC OSF/1, Digital UNIX)
2218 Win95/98/ME/2K/XP 2)
2220 z/OS (formerly OS/390)
2223 1) in DOS mode either the DOS or OS/2 ports can be used
2224 2) compilers: Borland, MinGW (GCC), VC6
2226 The following platforms worked with the previous releases (5.6 and
2227 5.7), but we did not manage either to fix or to test these in time
2228 for the 5.8.0 release. There is a very good chance that many of these
2229 will work fine with the 5.8.0.
2242 Known to be broken for 5.8.0 (but 5.6.1 and 5.7.2 can be used):
2246 The following platforms have been known to build Perl from source in
2247 the past (5.005_03 and earlier), but we haven't been able to verify
2248 their status for the current release, either because the
2249 hardware/software platforms are rare or because we don't have an
2250 active champion on these platforms--or both. They used to work,
2251 though, so go ahead and try compiling them, and let perlbug@perl.org
2284 The following platforms have their own source code distributions and
2285 binaries available via L<http://www.cpan.org/ports/>
2289 OS/400 (ILE) 5.005_02
2290 Tandem Guardian 5.004
2292 The following platforms have only binaries available via
2293 L<http://www.cpan.org/ports/index.html> :
2297 Acorn RISCOS 5.005_02
2301 Although we do suggest that you always build your own Perl from
2302 the source code, both for maximal configurability and for security,
2303 in case you are in a hurry you can check
2304 L<http://www.cpan.org/ports/index.html> for binary distributions.
2308 L<perlaix>, L<perlamiga>, L<perlbs2000>,
2309 L<perlce>, L<perlcygwin>, L<perldos>,
2310 L<perlebcdic>, L<perlfreebsd>, L<perlhurd>, L<perlhpux>, L<perlirix>,
2311 L<perlmacos>, L<perlmacosx>,
2312 L<perlnetware>, L<perlos2>, L<perlos390>, L<perlos400>,
2313 L<perlplan9>, L<perlqnx>, L<perlsolaris>, L<perltru64>,
2314 L<perlunicode>, L<perlvms>, L<perlvos>, L<perlwin32>, and L<Win32>.
2316 =head1 AUTHORS / CONTRIBUTORS
2318 Abigail <abigail@foad.org>,
2319 Charles Bailey <bailey@newman.upenn.edu>,
2320 Graham Barr <gbarr@pobox.com>,
2321 Tom Christiansen <tchrist@perl.com>,
2322 Nicholas Clark <nick@ccl4.org>,
2323 Thomas Dorner <Thomas.Dorner@start.de>,
2324 Andy Dougherty <doughera@lafayette.edu>,
2325 Dominic Dunlop <domo@computer.org>,
2326 Neale Ferguson <neale@vma.tabnsw.com.au>,
2327 David J. Fiander <davidf@mks.com>,
2328 Paul Green <Paul.Green@stratus.com>,
2329 M.J.T. Guy <mjtg@cam.ac.uk>,
2330 Jarkko Hietaniemi <jhi@iki.fi>,
2331 Luther Huffman <lutherh@stratcom.com>,
2332 Nick Ing-Simmons <nick@ing-simmons.net>,
2333 Andreas J. KE<ouml>nig <a.koenig@mind.de>,
2334 Markus Laker <mlaker@contax.co.uk>,
2335 Andrew M. Langmead <aml@world.std.com>,
2336 Larry Moore <ljmoore@freespace.net>,
2337 Paul Moore <Paul.Moore@uk.origin-it.com>,
2338 Chris Nandor <pudge@pobox.com>,
2339 Matthias Neeracher <neeracher@mac.com>,
2340 Philip Newton <pne@cpan.org>,
2341 Gary Ng <71564.1743@CompuServe.COM>,
2342 Tom Phoenix <rootbeer@teleport.com>,
2343 AndrE<eacute> Pirard <A.Pirard@ulg.ac.be>,
2344 Peter Prymmer <pvhp@forte.com>,
2345 Hugo van der Sanden <hv@crypt0.demon.co.uk>,
2346 Gurusamy Sarathy <gsar@activestate.com>,
2347 Paul J. Schinder <schinder@pobox.com>,
2348 Michael G Schwern <schwern@pobox.com>,
2349 Dan Sugalski <dan@sidhe.org>,
2350 Nathan Torkington <gnat@frii.com>,
2351 John Malmberg <wb8tyw@qsl.net>