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 The v-strings are portable only up to v2147483647 (0x7FFF_FFFF), that's
246 how far EBCDIC, or more precisely UTF-EBCDIC will go.
248 =head2 Files and Filesystems
250 Most platforms these days structure files in a hierarchical fashion.
251 So, it is reasonably safe to assume that all platforms support the
252 notion of a "path" to uniquely identify a file on the system. How
253 that path is really written, though, differs considerably.
255 Although similar, file path specifications differ between Unix,
256 Windows, S<Mac OS>, OS/2, VMS, VOS, S<RISC OS>, and probably others.
257 Unix, for example, is one of the few OSes that has the elegant idea
258 of a single root directory.
260 DOS, OS/2, VMS, VOS, and Windows can work similarly to Unix with C</>
261 as path separator, or in their own idiosyncratic ways (such as having
262 several root directories and various "unrooted" device files such NIL:
265 S<Mac OS> 9 and earlier used C<:> as a path separator instead of C</>.
267 The filesystem may support neither hard links (C<link>) nor
268 symbolic links (C<symlink>, C<readlink>, C<lstat>).
270 The filesystem may support neither access timestamp nor change
271 timestamp (meaning that about the only portable timestamp is the
272 modification timestamp), or one second granularity of any timestamps
273 (e.g. the FAT filesystem limits the time granularity to two seconds).
275 The "inode change timestamp" (the C<-C> filetest) may really be the
276 "creation timestamp" (which it is not in Unix).
278 VOS perl can emulate Unix filenames with C</> as path separator. The
279 native pathname characters greater-than, less-than, number-sign, and
280 percent-sign are always accepted.
282 S<RISC OS> perl can emulate Unix filenames with C</> as path
283 separator, or go native and use C<.> for path separator and C<:> to
284 signal filesystems and disk names.
286 Don't assume Unix filesystem access semantics: that read, write,
287 and execute are all the permissions there are, and even if they exist,
288 that their semantics (for example what do C<"r">, C<"w">, and C<"x"> mean on
289 a directory) are the Unix ones. The various Unix/POSIX compatibility
290 layers usually try to make interfaces like C<chmod()> work, but sometimes
291 there simply is no good mapping.
293 The C<File::Spec> modules provide methods to manipulate path
294 specifications and return the results in native format for each
295 platform. This is often unnecessary as Unix-style paths are
296 understood by Perl on every supported platform, but if you need to
297 produce native paths for a native utility that does not understand
298 Unix syntax, or if you are operating on paths or path components
299 in unknown (and thus possibly native) syntax, C<File::Spec> is
300 your friend. Here are two brief examples:
302 use File::Spec::Functions;
303 chdir(updir()); # go up one directory
305 # Concatenate a path from its components
306 my $file = catfile(updir(), 'temp', 'file.txt');
307 # on Unix: '../temp/file.txt'
308 # on Win32: '..\temp\file.txt'
309 # on VMS: '[-.temp]file.txt'
311 In general, production code should not have file paths hardcoded.
312 Making them user-supplied or read from a configuration file is
313 better, keeping in mind that file path syntax varies on different
316 This is especially noticeable in scripts like Makefiles and test suites,
317 which often assume C</> as a path separator for subdirectories.
319 Also of use is C<File::Basename> from the standard distribution, which
320 splits a pathname into pieces (base filename, full path to directory,
323 Even when on a single platform (if you can call Unix a single platform),
324 remember not to count on the existence or the contents of particular
325 system-specific files or directories, like F</etc/passwd>,
326 F</etc/sendmail.conf>, F</etc/resolv.conf>, or even F</tmp/>. For
327 example, F</etc/passwd> may exist but not contain the encrypted
328 passwords, because the system is using some form of enhanced security.
329 Or it may not contain all the accounts, because the system is using NIS.
330 If code does need to rely on such a file, include a description of the
331 file and its format in the code's documentation, then make it easy for
332 the user to override the default location of the file.
334 Don't assume a text file will end with a newline. They should,
337 Do not have two files or directories of the same name with different
338 case, like F<test.pl> and F<Test.pl>, as many platforms have
339 case-insensitive (or at least case-forgiving) filenames. Also, try
340 not to have non-word characters (except for C<.>) in the names, and
341 keep them to the 8.3 convention, for maximum portability, onerous a
342 burden though this may appear.
344 Likewise, when using the C<AutoSplit> module, try to keep your functions to
345 8.3 naming and case-insensitive conventions; or, at the least,
346 make it so the resulting files have a unique (case-insensitively)
349 Whitespace in filenames is tolerated on most systems, but not all,
350 and even on systems where it might be tolerated, some utilities
351 might become confused by such whitespace.
353 Many systems (DOS, VMS ODS-2) cannot have more than one C<.> in their
356 Don't assume C<< > >> won't be the first character of a filename.
357 Always use C<< < >> explicitly to open a file for reading, or even
358 better, use the three-arg version of C<open>, unless you want the user to
359 be able to specify a pipe open.
361 open my $fh, '<', $existing_file) or die $!;
363 If filenames might use strange characters, it is safest to open it
364 with C<sysopen> instead of C<open>. C<open> is magic and can
365 translate characters like C<< > >>, C<< < >>, and C<|>, which may
366 be the wrong thing to do. (Sometimes, though, it's the right thing.)
367 Three-arg open can also help protect against this translation in cases
368 where it is undesirable.
370 Don't use C<:> as a part of a filename since many systems use that for
371 their own semantics (Mac OS Classic for separating pathname components,
372 many networking schemes and utilities for separating the nodename and
373 the pathname, and so on). For the same reasons, avoid C<@>, C<;> and
376 Don't assume that in pathnames you can collapse two leading slashes
377 C<//> into one: some networking and clustering filesystems have special
378 semantics for that. Let the operating system sort it out.
380 The I<portable filename characters> as defined by ANSI C are
382 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
383 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
387 and the C<"-"> shouldn't be the first character. If you want to be
388 hypercorrect, stay case-insensitive and within the 8.3 naming
389 convention (all the files and directories have to be unique within one
390 directory if their names are lowercased and truncated to eight
391 characters before the C<.>, if any, and to three characters after the
392 C<.>, if any). (And do not use C<.>s in directory names.)
394 =head2 System Interaction
396 Not all platforms provide a command line. These are usually platforms
397 that rely primarily on a Graphical User Interface (GUI) for user
398 interaction. A program requiring a command line interface might
399 not work everywhere. This is probably for the user of the program
400 to deal with, so don't stay up late worrying about it.
402 Some platforms can't delete or rename files held open by the system,
403 this limitation may also apply to changing filesystem metainformation
404 like file permissions or owners. Remember to C<close> files when you
405 are done with them. Don't C<unlink> or C<rename> an open file. Don't
406 C<tie> or C<open> a file already tied or opened; C<untie> or C<close>
409 Don't open the same file more than once at a time for writing, as some
410 operating systems put mandatory locks on such files.
412 Don't assume that write/modify permission on a directory gives the
413 right to add or delete files/directories in that directory. That is
414 filesystem specific: in some filesystems you need write/modify
415 permission also (or even just) in the file/directory itself. In some
416 filesystems (AFS, DFS) the permission to add/delete directory entries
417 is a completely separate permission.
419 Don't assume that a single C<unlink> completely gets rid of the file:
420 some filesystems (most notably the ones in VMS) have versioned
421 filesystems, and C<unlink()> removes only the most recent one (it doesn't
422 remove all the versions because by default the native tools on those
423 platforms remove just the most recent version, too). The portable
424 idiom to remove all the versions of a file is
426 1 while unlink "file";
428 This will terminate if the file is undeleteable for some reason
429 (protected, not there, and so on).
431 Don't count on a specific environment variable existing in C<%ENV>.
432 Don't count on C<%ENV> entries being case-sensitive, or even
433 case-preserving. Don't try to clear C<%ENV> by saying C<%ENV = ();>, or,
434 if you really have to, make it conditional on C<$^O ne 'VMS'> since in
435 VMS the C<%ENV> table is much more than a per-process key-value string
438 On VMS, some entries in the C<%ENV> hash are dynamically created when
439 their key is used on a read if they did not previously exist. The
440 values for C<$ENV{HOME}>, C<$ENV{TERM}>, C<$ENV{PATH}>, and C<$ENV{USER}>,
441 are known to be dynamically generated. The specific names that are
442 dynamically generated may vary with the version of the C library on VMS,
443 and more may exist than are documented.
445 On VMS by default, changes to the %ENV hash persist after perl exits.
446 Subsequent invocations of perl in the same process can inadvertently
447 inherit environment settings that were meant to be temporary.
449 Don't count on signals or C<%SIG> for anything.
451 Don't count on filename globbing. Use C<opendir>, C<readdir>, and
454 Don't count on per-program environment variables, or per-program current
457 Don't count on specific values of C<$!>, neither numeric nor
458 especially the string values. Users may switch their locales causing
459 error messages to be translated into their languages. If you can
460 trust a POSIXish environment, you can portably use the symbols defined
461 by the C<Errno> module, like C<ENOENT>. And don't trust on the values of C<$!>
462 at all except immediately after a failed system call.
464 =head2 Command names versus file pathnames
466 Don't assume that the name used to invoke a command or program with
467 C<system> or C<exec> can also be used to test for the existence of the
468 file that holds the executable code for that command or program.
469 First, many systems have "internal" commands that are built-in to the
470 shell or OS and while these commands can be invoked, there is no
471 corresponding file. Second, some operating systems (e.g., Cygwin,
472 DJGPP, OS/2, and VOS) have required suffixes for executable files;
473 these suffixes are generally permitted on the command name but are not
474 required. Thus, a command like F<"perl"> might exist in a file named
475 F<"perl">, F<"perl.exe">, or F<"perl.pm">, depending on the operating system.
476 The variable C<"_exe"> in the C<Config> module holds the executable suffix,
477 if any. Third, the VMS port carefully sets up C<$^X> and
478 C<$Config{perlpath}> so that no further processing is required. This is
479 just as well, because the matching regular expression used below would
480 then have to deal with a possible trailing version number in the VMS
483 To convert C<$^X> to a file pathname, taking account of the requirements
484 of the various operating system possibilities, say:
489 {$thisperl .= $Config{_exe} unless $thisperl =~ m/$Config{_exe}$/i;}
491 To convert C<$Config{perlpath}> to a file pathname, say:
494 my $thisperl = $Config{perlpath};
496 {$thisperl .= $Config{_exe} unless $thisperl =~ m/$Config{_exe}$/i;}
500 Don't assume that you can reach the public Internet.
502 Don't assume that there is only one way to get through firewalls
503 to the public Internet.
505 Don't assume that you can reach outside world through any other port
506 than 80, or some web proxy. ftp is blocked by many firewalls.
508 Don't assume that you can send email by connecting to the local SMTP port.
510 Don't assume that you can reach yourself or any node by the name
511 'localhost'. The same goes for '127.0.0.1'. You will have to try both.
513 Don't assume that the host has only one network card, or that it
514 can't bind to many virtual IP addresses.
516 Don't assume a particular network device name.
518 Don't assume a particular set of C<ioctl()>s will work.
520 Don't assume that you can ping hosts and get replies.
522 Don't assume that any particular port (service) will respond.
524 Don't assume that C<Sys::Hostname> (or any other API or command) returns
525 either a fully qualified hostname or a non-qualified hostname: it all
526 depends on how the system had been configured. Also remember that for
527 things such as DHCP and NAT, the hostname you get back might not be
530 All the above "don't":s may look daunting, and they are, but the key
531 is to degrade gracefully if one cannot reach the particular network
532 service one wants. Croaking or hanging do not look very professional.
534 =head2 Interprocess Communication (IPC)
536 In general, don't directly access the system in code meant to be
537 portable. That means, no C<system>, C<exec>, C<fork>, C<pipe>,
538 C<``>, C<qx//>, C<open> with a C<|>, nor any of the other things
539 that makes being a Perl hacker worth being.
541 Commands that launch external processes are generally supported on
542 most platforms (though many of them do not support any type of
543 forking). The problem with using them arises from what you invoke
544 them on. External tools are often named differently on different
545 platforms, may not be available in the same location, might accept
546 different arguments, can behave differently, and often present their
547 results in a platform-dependent way. Thus, you should seldom depend
548 on them to produce consistent results. (Then again, if you're calling
549 I<netstat -a>, you probably don't expect it to run on both Unix and CP/M.)
551 One especially common bit of Perl code is opening a pipe to B<sendmail>:
553 open(MAIL, '|/usr/lib/sendmail -t')
554 or die "cannot fork sendmail: $!";
556 This is fine for systems programming when sendmail is known to be
557 available. But it is not fine for many non-Unix systems, and even
558 some Unix systems that may not have sendmail installed. If a portable
559 solution is needed, see the various distributions on CPAN that deal
560 with it. C<Mail::Mailer> and C<Mail::Send> in the C<MailTools> distribution are
561 commonly used, and provide several mailing methods, including C<mail>,
562 C<sendmail>, and direct SMTP (via C<Net::SMTP>) if a mail transfer agent is
563 not available. C<Mail::Sendmail> is a standalone module that provides
564 simple, platform-independent mailing.
566 The Unix System V IPC (C<msg*(), sem*(), shm*()>) is not available
567 even on all Unix platforms.
569 Do not use either the bare result of C<pack("N", 10, 20, 30, 40)> or
570 bare v-strings (such as C<v10.20.30.40>) to represent IPv4 addresses:
571 both forms just pack the four bytes into network order. That this
572 would be equal to the C language C<in_addr> struct (which is what the
573 socket code internally uses) is not guaranteed. To be portable use
574 the routines of the C<Socket> extension, such as C<inet_aton()>,
575 C<inet_ntoa()>, and C<sockaddr_in()>.
577 The rule of thumb for portable code is: Do it all in portable Perl, or
578 use a module (that may internally implement it with platform-specific
579 code, but exposes a common interface).
581 =head2 External Subroutines (XS)
583 XS code can usually be made to work with any platform, but dependent
584 libraries, header files, etc., might not be readily available or
585 portable, or the XS code itself might be platform-specific, just as Perl
586 code might be. If the libraries and headers are portable, then it is
587 normally reasonable to make sure the XS code is portable, too.
589 A different type of portability issue arises when writing XS code:
590 availability of a C compiler on the end-user's system. C brings
591 with it its own portability issues, and writing XS code will expose
592 you to some of those. Writing purely in Perl is an easier way to
595 =head2 Standard Modules
597 In general, the standard modules work across platforms. Notable
598 exceptions are the C<CPAN> module (which currently makes connections to external
599 programs that may not be available), platform-specific modules (like
600 C<ExtUtils::MM_VMS>), and DBM modules.
602 There is no one DBM module available on all platforms.
603 C<SDBM_File> and the others are generally available on all Unix and DOSish
604 ports, but not in MacPerl, where only C<NDBM_File> and C<DB_File> are
607 The good news is that at least some DBM module should be available, and
608 C<AnyDBM_File> will use whichever module it can find. Of course, then
609 the code needs to be fairly strict, dropping to the greatest common
610 factor (e.g., not exceeding 1K for each record), so that it will
611 work with any DBM module. See L<AnyDBM_File> for more details.
615 The system's notion of time of day and calendar date is controlled in
616 widely different ways. Don't assume the timezone is stored in C<$ENV{TZ}>,
617 and even if it is, don't assume that you can control the timezone through
618 that variable. Don't assume anything about the three-letter timezone
619 abbreviations (for example that MST would be the Mountain Standard Time,
620 it's been known to stand for Moscow Standard Time). If you need to
621 use timezones, express them in some unambiguous format like the
622 exact number of minutes offset from UTC, or the POSIX timezone
625 Don't assume that the epoch starts at 00:00:00, January 1, 1970,
626 because that is OS- and implementation-specific. It is better to
627 store a date in an unambiguous representation. The ISO 8601 standard
628 defines YYYY-MM-DD as the date format, or YYYY-MM-DDTHH:MM:SS
629 (that's a literal "T" separating the date from the time).
630 Please do use the ISO 8601 instead of making us guess what
631 date 02/03/04 might be. ISO 8601 even sorts nicely as-is.
632 A text representation (like "1987-12-18") can be easily converted
633 into an OS-specific value using a module like C<Date::Parse>.
634 An array of values, such as those returned by C<localtime>, can be
635 converted to an OS-specific representation using C<Time::Local>.
637 When calculating specific times, such as for tests in time or date modules,
638 it may be appropriate to calculate an offset for the epoch.
641 my $offset = Time::Local::timegm(0, 0, 0, 1, 0, 70);
643 The value for C<$offset> in Unix will be C<0>, but in Mac OS Classic
644 will be some large number. C<$offset> can then be added to a Unix time
645 value to get what should be the proper value on any system.
647 =head2 Character sets and character encoding
649 Assume very little about character sets.
651 Assume nothing about numerical values (C<ord>, C<chr>) of characters.
652 Do not use explicit code point ranges (like C<\xHH-\xHH)>. However,
653 starting in Perl v5.22, regular expression pattern bracketed character
654 class ranges specified like C<qr/[\N{U+HH}-\N{U+HH}]/> are portable.
655 You can portably use symbolic character classes like C<[:print:]>.
657 Do not assume that the alphabetic characters are encoded contiguously
658 (in the numeric sense). There may be gaps. Special coding in Perl,
659 however, guarantees that all subsets of C<qr/[A-Z]/>, C<qr/[a-z]/>, and
660 C<qr/[0-9]/> behave as expected. C<tr///> behaves the same for these
661 ranges. In patterns, any ranges specified with end points using the
662 C<\N{...}> notations ensures character set portability, but it is a bug
663 in Perl v5.22, that this isn't true of C<tr///>.
665 Do not assume anything about the ordering of the characters.
666 The lowercase letters may come before or after the uppercase letters;
667 the lowercase and uppercase may be interlaced so that both "a" and "A"
668 come before "b"; the accented and other international characters may
669 be interlaced so that E<auml> comes before "b".
670 L<Unicode::Collate> can be used to sort this all out.
672 =head2 Internationalisation
674 If you may assume POSIX (a rather large assumption), you may read
675 more about the POSIX locale system from L<perllocale>. The locale
676 system at least attempts to make things a little bit more portable,
677 or at least more convenient and native-friendly for non-English
678 users. The system affects character sets and encoding, and date
679 and time formatting--amongst other things.
681 If you really want to be international, you should consider Unicode.
682 See L<perluniintro> and L<perlunicode> for more information.
684 If you want to use non-ASCII bytes (outside the bytes 0x00..0x7f) in
685 the "source code" of your code, to be portable you have to be explicit
686 about what bytes they are. Someone might for example be using your
687 code under a UTF-8 locale, in which case random native bytes might be
688 illegal ("Malformed UTF-8 ...") This means that for example embedding
689 ISO 8859-1 bytes beyond 0x7f into your strings might cause trouble
690 later. If the bytes are native 8-bit bytes, you can use the C<bytes>
691 pragma. If the bytes are in a string (regular expressions being
692 curious strings), you can often also use the C<\xHH> or more portably,
693 the C<\N{U+HH}> notations instead
694 of embedding the bytes as-is. If you want to write your code in UTF-8,
697 =head2 System Resources
699 If your code is destined for systems with severely constrained (or
700 missing!) virtual memory systems then you want to be I<especially> mindful
701 of avoiding wasteful constructs such as:
703 my @lines = <$very_large_file>; # bad
705 while (<$fh>) {$file .= $_} # sometimes bad
706 my $file = join('', <$fh>); # better
708 The last two constructs may appear unintuitive to most people. The
709 first repeatedly grows a string, whereas the second allocates a
710 large chunk of memory in one go. On some systems, the second is
711 more efficient than the first.
715 Most multi-user platforms provide basic levels of security, usually
716 implemented at the filesystem level. Some, however, unfortunately do
717 not. Thus the notion of user id, or "home" directory,
718 or even the state of being logged-in, may be unrecognizable on many
719 platforms. If you write programs that are security-conscious, it
720 is usually best to know what type of system you will be running
721 under so that you can write code explicitly for that platform (or
724 Don't assume the Unix filesystem access semantics: the operating
725 system or the filesystem may be using some ACL systems, which are
726 richer languages than the usual C<rwx>. Even if the C<rwx> exist,
727 their semantics might be different.
729 (From the security viewpoint, testing for permissions before attempting to
730 do something is silly anyway: if one tries this, there is potential
731 for race conditions. Someone or something might change the
732 permissions between the permissions check and the actual operation.
733 Just try the operation.)
735 Don't assume the Unix user and group semantics: especially, don't
736 expect C<< $< >> and C<< $> >> (or C<$(> and C<$)>) to work
737 for switching identities (or memberships).
739 Don't assume set-uid and set-gid semantics. (And even if you do,
740 think twice: set-uid and set-gid are a known can of security worms.)
744 For those times when it is necessary to have platform-specific code,
745 consider keeping the platform-specific code in one place, making porting
746 to other platforms easier. Use the C<Config> module and the special
747 variable C<$^O> to differentiate platforms, as described in
750 Be careful in the tests you supply with your module or programs.
751 Module code may be fully portable, but its tests might not be. This
752 often happens when tests spawn off other processes or call external
753 programs to aid in the testing, or when (as noted above) the tests
754 assume certain things about the filesystem and paths. Be careful not
755 to depend on a specific output style for errors, such as when checking
756 C<$!> after a failed system call. Using C<$!> for anything else than
757 displaying it as output is doubtful (though see the C<Errno> module for
758 testing reasonably portably for error value). Some platforms expect
759 a certain output format, and Perl on those platforms may have been
760 adjusted accordingly. Most specifically, don't anchor a regex when
761 testing an error value.
765 Modules uploaded to CPAN are tested by a variety of volunteers on
766 different platforms. These CPAN testers are notified by mail of each
767 new upload, and reply to the list with PASS, FAIL, NA (not applicable to
768 this platform), or UNKNOWN (unknown), along with any relevant notations.
770 The purpose of the testing is twofold: one, to help developers fix any
771 problems in their code that crop up because of lack of testing on other
772 platforms; two, to provide users with information about whether
773 a given module works on a given platform.
781 Mailing list: cpan-testers-discuss@perl.org
785 Testing results: L<http://www.cpantesters.org/>
791 Perl is built with a C<$^O> variable that indicates the operating
792 system it was built on. This was implemented
793 to help speed up code that would otherwise have to C<use Config>
794 and use the value of C<$Config{osname}>. Of course, to get more
795 detailed information about the system, looking into C<%Config> is
796 certainly recommended.
798 C<%Config> cannot always be trusted, however, because it was built
799 at compile time. If perl was built in one place, then transferred
800 elsewhere, some values may be wrong. The values may even have been
801 edited after the fact.
805 Perl works on a bewildering variety of Unix and Unix-like platforms (see
806 e.g. most of the files in the F<hints/> directory in the source code kit).
807 On most of these systems, the value of C<$^O> (hence C<$Config{'osname'}>,
808 too) is determined either by lowercasing and stripping punctuation from the
809 first field of the string returned by typing C<uname -a> (or a similar command)
810 at the shell prompt or by testing the file system for the presence of
811 uniquely named files such as a kernel or header file. Here, for example,
812 are a few of the more popular Unix flavors:
814 uname $^O $Config{'archname'}
815 --------------------------------------------
817 BSD/OS bsdos i386-bsdos
819 DYNIX/ptx dynixptx i386-dynixptx
820 FreeBSD freebsd freebsd-i386
821 Haiku haiku BePC-haiku
822 Linux linux arm-linux
823 Linux linux armv5tel-linux
824 Linux linux i386-linux
825 Linux linux i586-linux
826 Linux linux ppc-linux
827 HP-UX hpux PA-RISC1.1
829 Mac OS X darwin darwin
831 NeXT 4 next OPENSTEP-Mach
832 openbsd openbsd i386-openbsd
833 OSF1 dec_osf alpha-dec_osf
834 reliantunix-n svr4 RM400-svr4
835 SCO_SV sco_sv i386-sco_sv
836 SINIX-N svr4 RM400-svr4
837 sn4609 unicos CRAY_C90-unicos
838 sn6521 unicosmk t3e-unicosmk
839 sn9617 unicos CRAY_J90-unicos
840 SunOS solaris sun4-solaris
841 SunOS solaris i86pc-solaris
842 SunOS4 sunos sun4-sunos
844 Because the value of C<$Config{archname}> may depend on the
845 hardware architecture, it can vary more than the value of C<$^O>.
847 =head2 DOS and Derivatives
849 Perl has long been ported to Intel-style microcomputers running under
850 systems like PC-DOS, MS-DOS, OS/2, and most Windows platforms you can
851 bring yourself to mention (except for Windows CE, if you count that).
852 Users familiar with I<COMMAND.COM> or I<CMD.EXE> style shells should
853 be aware that each of these file specifications may have subtle
856 my $filespec0 = "c:/foo/bar/file.txt";
857 my $filespec1 = "c:\\foo\\bar\\file.txt";
858 my $filespec2 = 'c:\foo\bar\file.txt';
859 my $filespec3 = 'c:\\foo\\bar\\file.txt';
861 System calls accept either C</> or C<\> as the path separator.
862 However, many command-line utilities of DOS vintage treat C</> as
863 the option prefix, so may get confused by filenames containing C</>.
864 Aside from calling any external programs, C</> will work just fine,
865 and probably better, as it is more consistent with popular usage,
866 and avoids the problem of remembering what to backwhack and what
869 The DOS FAT filesystem can accommodate only "8.3" style filenames. Under
870 the "case-insensitive, but case-preserving" HPFS (OS/2) and NTFS (NT)
871 filesystems you may have to be careful about case returned with functions
872 like C<readdir> or used with functions like C<open> or C<opendir>.
874 DOS also treats several filenames as special, such as AUX, PRN,
875 NUL, CON, COM1, LPT1, LPT2, etc. Unfortunately, sometimes these
876 filenames won't even work if you include an explicit directory
877 prefix. It is best to avoid such filenames, if you want your code
878 to be portable to DOS and its derivatives. It's hard to know what
879 these all are, unfortunately.
881 Users of these operating systems may also wish to make use of
882 scripts such as I<pl2bat.bat> or I<pl2cmd> to
883 put wrappers around your scripts.
885 Newline (C<\n>) is translated as C<\015\012> by STDIO when reading from
886 and writing to files (see L<"Newlines">). C<binmode(FILEHANDLE)>
887 will keep C<\n> translated as C<\012> for that filehandle. Since it is a
888 no-op on other systems, C<binmode> should be used for cross-platform code
889 that deals with binary data. That's assuming you realize in advance
890 that your data is in binary. General-purpose programs should
891 often assume nothing about their data.
893 The C<$^O> variable and the C<$Config{archname}> values for various
894 DOSish perls are as follows:
896 OS $^O $Config{archname} ID Version
897 --------------------------------------------------------
901 Windows 3.1 ? ? 0 3 01
902 Windows 95 MSWin32 MSWin32-x86 1 4 00
903 Windows 98 MSWin32 MSWin32-x86 1 4 10
904 Windows ME MSWin32 MSWin32-x86 1 ?
905 Windows NT MSWin32 MSWin32-x86 2 4 xx
906 Windows NT MSWin32 MSWin32-ALPHA 2 4 xx
907 Windows NT MSWin32 MSWin32-ppc 2 4 xx
908 Windows 2000 MSWin32 MSWin32-x86 2 5 00
909 Windows XP MSWin32 MSWin32-x86 2 5 01
910 Windows 2003 MSWin32 MSWin32-x86 2 5 02
911 Windows Vista MSWin32 MSWin32-x86 2 6 00
912 Windows 7 MSWin32 MSWin32-x86 2 6 01
913 Windows 7 MSWin32 MSWin32-x64 2 6 01
914 Windows 2008 MSWin32 MSWin32-x86 2 6 01
915 Windows 2008 MSWin32 MSWin32-x64 2 6 01
916 Windows CE MSWin32 ? 3
919 The various MSWin32 Perl's can distinguish the OS they are running on
920 via the value of the fifth element of the list returned from
921 C<Win32::GetOSVersion()>. For example:
923 if ($^O eq 'MSWin32') {
924 my @os_version_info = Win32::GetOSVersion();
925 print +('3.1','95','NT')[$os_version_info[4]],"\n";
928 There are also C<Win32::IsWinNT()> and C<Win32::IsWin95()>; try C<perldoc Win32>,
929 and as of libwin32 0.19 (not part of the core Perl distribution)
930 C<Win32::GetOSName()>. The very portable C<POSIX::uname()> will work too:
932 c:\> perl -MPOSIX -we "print join '|', uname"
933 Windows NT|moonru|5.0|Build 2195 (Service Pack 2)|x86
941 The djgpp environment for DOS, L<http://www.delorie.com/djgpp/>
946 The EMX environment for DOS, OS/2, etc. emx@iaehv.nl,
947 L<ftp://hobbes.nmsu.edu/pub/os2/dev/emx/> Also L<perlos2>.
951 Build instructions for Win32 in L<perlwin32>, or under the Cygnus environment
956 The C<Win32::*> modules in L<Win32>.
960 The ActiveState Pages, L<http://www.activestate.com/>
964 The Cygwin environment for Win32; F<README.cygwin> (installed
965 as L<perlcygwin>), L<http://www.cygwin.com/>
969 The U/WIN environment for Win32,
970 L<http://www.research.att.com/sw/tools/uwin/>
974 Build instructions for OS/2, L<perlos2>
980 Perl on VMS is discussed in L<perlvms> in the Perl distribution.
982 The official name of VMS as of this writing is OpenVMS.
984 Interacting with Perl from the Digital Command Language (DCL) shell
985 often requires a different set of quotation marks than Unix shells do.
988 $ perl -e "print ""Hello, world.\n"""
991 There are several ways to wrap your Perl scripts in DCL F<.COM> files, if
992 you are so inclined. For example:
994 $ write sys$output "Hello from DCL!"
996 $ then perl -x 'f$environment("PROCEDURE")
997 $ else perl -x - 'p1 'p2 'p3 'p4 'p5 'p6 'p7 'p8
998 $ deck/dollars="__END__"
1001 print "Hello from Perl!\n";
1006 Do take care with C<$ ASSIGN/nolog/user SYS$COMMAND: SYS$INPUT> if your
1007 Perl-in-DCL script expects to do things like C<< $read = <STDIN>; >>.
1009 The VMS operating system has two filesystems, designated by their
1010 on-disk structure (ODS) level: ODS-2 and its successor ODS-5. The
1011 initial port of Perl to VMS pre-dates ODS-5, but all current testing and
1012 development assumes ODS-5 and its capabilities, including case
1013 preservation, extended characters in filespecs, and names up to 8192
1016 Perl on VMS can accept either VMS- or Unix-style file
1017 specifications as in either of the following:
1019 $ perl -ne "print if /perl_setup/i" SYS$LOGIN:LOGIN.COM
1020 $ perl -ne "print if /perl_setup/i" /sys$login/login.com
1022 but not a mixture of both as in:
1024 $ perl -ne "print if /perl_setup/i" sys$login:/login.com
1025 Can't open sys$login:/login.com: file specification syntax error
1027 In general, the easiest path to portability is always to specify
1028 filenames in Unix format unless they will need to be processed by native
1029 commands or utilities. Because of this latter consideration, the
1030 File::Spec module by default returns native format specifications
1031 regardless of input format. This default may be reversed so that
1032 filenames are always reported in Unix format by specifying the
1033 C<DECC$FILENAME_UNIX_REPORT> feature logical in the environment.
1035 The file type, or extension, is always present in a VMS-format file
1036 specification even if it's zero-length. This means that, by default,
1037 C<readdir> will return a trailing dot on a file with no extension, so
1038 where you would see C<"a"> on Unix you'll see C<"a."> on VMS. However,
1039 the trailing dot may be suppressed by enabling the
1040 C<DECC$READDIR_DROPDOTNOTYPE> feature in the environment (see the CRTL
1041 documentation on feature logical names).
1043 What C<\n> represents depends on the type of file opened. It usually
1044 represents C<\012> but it could also be C<\015>, C<\012>, C<\015\012>,
1045 C<\000>, C<\040>, or nothing depending on the file organization and
1046 record format. The C<VMS::Stdio> module provides access to the
1047 special C<fopen()> requirements of files with unusual attributes on VMS.
1049 The value of C<$^O> on OpenVMS is "VMS". To determine the architecture
1050 that you are running on refer to C<$Config{'archname'}>.
1052 On VMS, perl determines the UTC offset from the C<SYS$TIMEZONE_DIFFERENTIAL>
1053 logical name. Although the VMS epoch began at 17-NOV-1858 00:00:00.00,
1054 calls to C<localtime> are adjusted to count offsets from
1055 01-JAN-1970 00:00:00.00, just like Unix.
1063 F<README.vms> (installed as F<README_vms>), L<perlvms>
1067 vmsperl list, vmsperl-subscribe@perl.org
1071 vmsperl on the web, L<http://www.sidhe.org/vmsperl/index.html>
1075 VMS Software Inc. web site, L<http://www.vmssoftware.com>
1081 Perl on VOS (also known as OpenVOS) is discussed in F<README.vos>
1082 in the Perl distribution (installed as L<perlvos>). Perl on VOS
1083 can accept either VOS- or Unix-style file specifications as in
1084 either of the following:
1086 $ perl -ne "print if /perl_setup/i" >system>notices
1087 $ perl -ne "print if /perl_setup/i" /system/notices
1089 or even a mixture of both as in:
1091 $ perl -ne "print if /perl_setup/i" >system/notices
1093 Even though VOS allows the slash character to appear in object
1094 names, because the VOS port of Perl interprets it as a pathname
1095 delimiting character, VOS files, directories, or links whose
1096 names contain a slash character cannot be processed. Such files
1097 must be renamed before they can be processed by Perl.
1099 Older releases of VOS (prior to OpenVOS Release 17.0) limit file
1100 names to 32 or fewer characters, prohibit file names from
1101 starting with a C<-> character, and prohibit file names from
1102 containing any character matching C<< tr/ !#%&'()*;<=>?// >>.
1104 Newer releases of VOS (OpenVOS Release 17.0 or later) support a
1105 feature known as extended names. On these releases, file names
1106 can contain up to 255 characters, are prohibited from starting
1107 with a C<-> character, and the set of prohibited characters is
1108 reduced to any character matching C<< tr/#%*<>?// >>. There are
1109 restrictions involving spaces and apostrophes: these characters
1110 must not begin or end a name, nor can they immediately precede or
1111 follow a period. Additionally, a space must not immediately
1112 precede another space or hyphen. Specifically, the following
1113 character combinations are prohibited: space-space,
1114 space-hyphen, period-space, space-period, period-apostrophe,
1115 apostrophe-period, leading or trailing space, and leading or
1116 trailing apostrophe. Although an extended file name is limited
1117 to 255 characters, a path name is still limited to 256
1120 The value of C<$^O> on VOS is "vos". To determine the
1121 architecture that you are running on without resorting to loading
1122 all of C<%Config> you can examine the content of the C<@INC> array
1126 print "I'm on a Stratus box!\n";
1128 print "I'm not on a Stratus box!\n";
1138 F<README.vos> (installed as L<perlvos>)
1142 The VOS mailing list.
1144 There is no specific mailing list for Perl on VOS. You can contact
1145 the Stratus Technologies Customer Assistance Center (CAC) for your
1146 region, or you can use the contact information located in the
1147 distribution files on the Stratus Anonymous FTP site.
1151 Stratus Technologies on the web at L<http://www.stratus.com>
1155 VOS Open-Source Software on the web at L<http://ftp.stratus.com/pub/vos/vos.html>
1159 =head2 EBCDIC Platforms
1161 v5.22 core Perl runs on z/OS (formerly OS/390). Theoretically it could
1162 run on the successors of OS/400 on AS/400 minicomputers as well as
1163 VM/ESA, and BS2000 for S/390 Mainframes. Such computers use EBCDIC
1164 character sets internally (usually
1165 Character Code Set ID 0037 for OS/400 and either 1047 or POSIX-BC for S/390
1168 The rest of this section may need updating, but we don't know what it
1169 should say. Please email comments to
1170 L<perlbug@perl.org|mailto:perlbug@perl.org>.
1172 On the mainframe Perl currently works under the "Unix system
1173 services for OS/390" (formerly known as OpenEdition), VM/ESA OpenEdition, or
1174 the BS200 POSIX-BC system (BS2000 is supported in Perl 5.6 and greater).
1175 See L<perlos390> for details. Note that for OS/400 there is also a port of
1176 Perl 5.8.1/5.10.0 or later to the PASE which is ASCII-based (as opposed to
1177 ILE which is EBCDIC-based), see L<perlos400>.
1179 As of R2.5 of USS for OS/390 and Version 2.3 of VM/ESA these Unix
1180 sub-systems do not support the C<#!> shebang trick for script invocation.
1181 Hence, on OS/390 and VM/ESA Perl scripts can be executed with a header
1182 similar to the following simple script:
1185 eval 'exec /usr/local/bin/perl -S $0 ${1+"$@"}'
1187 #!/usr/local/bin/perl # just a comment really
1189 print "Hello from perl!\n";
1191 OS/390 will support the C<#!> shebang trick in release 2.8 and beyond.
1192 Calls to C<system> and backticks can use POSIX shell syntax on all
1195 On the AS/400, if PERL5 is in your library list, you may need
1196 to wrap your Perl scripts in a CL procedure to invoke them like so:
1199 CALL PGM(PERL5/PERL) PARM('/QOpenSys/hello.pl')
1202 This will invoke the Perl script F<hello.pl> in the root of the
1203 QOpenSys file system. On the AS/400 calls to C<system> or backticks
1206 On these platforms, bear in mind that the EBCDIC character set may have
1207 an effect on what happens with some Perl functions (such as C<chr>,
1208 C<pack>, C<print>, C<printf>, C<ord>, C<sort>, C<sprintf>, C<unpack>), as
1209 well as bit-fiddling with ASCII constants using operators like C<^>, C<&>
1210 and C<|>, not to mention dealing with socket interfaces to ASCII computers
1211 (see L<"Newlines">).
1213 Fortunately, most web servers for the mainframe will correctly
1214 translate the C<\n> in the following statement to its ASCII equivalent
1215 (C<\r> is the same under both Unix and z/OS):
1217 print "Content-type: text/html\r\n\r\n";
1219 The values of C<$^O> on some of these platforms includes:
1221 uname $^O $Config{'archname'}
1222 --------------------------------------------
1225 POSIX-BC posix-bc BS2000-posix-bc
1227 Some simple tricks for determining if you are running on an EBCDIC
1228 platform could include any of the following (perhaps all):
1230 if ("\t" eq "\005") { print "EBCDIC may be spoken here!\n"; }
1232 if (ord('A') == 193) { print "EBCDIC may be spoken here!\n"; }
1234 if (chr(169) eq 'z') { print "EBCDIC may be spoken here!\n"; }
1236 One thing you may not want to rely on is the EBCDIC encoding
1237 of punctuation characters since these may differ from code page to code
1238 page (and once your module or script is rumoured to work with EBCDIC,
1239 folks will want it to work with all EBCDIC character sets).
1247 L<perlos390>, L<perlos400>, L<perlbs2000>, L<perlebcdic>.
1251 The perl-mvs@perl.org list is for discussion of porting issues as well as
1252 general usage issues for all EBCDIC Perls. Send a message body of
1253 "subscribe perl-mvs" to majordomo@perl.org.
1257 AS/400 Perl information at
1258 L<http://as400.rochester.ibm.com/>
1259 as well as on CPAN in the F<ports/> directory.
1263 =head2 Acorn RISC OS
1265 Because Acorns use ASCII with newlines (C<\n>) in text files as C<\012> like
1266 Unix, and because Unix filename emulation is turned on by default,
1267 most simple scripts will probably work "out of the box". The native
1268 filesystem is modular, and individual filesystems are free to be
1269 case-sensitive or insensitive, and are usually case-preserving. Some
1270 native filesystems have name length limits, which file and directory
1271 names are silently truncated to fit. Scripts should be aware that the
1272 standard filesystem currently has a name length limit of B<10>
1273 characters, with up to 77 items in a directory, but other filesystems
1274 may not impose such limitations.
1276 Native filenames are of the form
1278 Filesystem#Special_Field::DiskName.$.Directory.Directory.File
1282 Special_Field is not usually present, but may contain . and $ .
1283 Filesystem =~ m|[A-Za-z0-9_]|
1284 DsicName =~ m|[A-Za-z0-9_/]|
1285 $ represents the root directory
1286 . is the path separator
1287 @ is the current directory (per filesystem but machine global)
1288 ^ is the parent directory
1289 Directory and File =~ m|[^\0- "\.\$\%\&:\@\\^\|\177]+|
1291 The default filename translation is roughly C<tr|/.|./|;>
1293 Note that C<"ADFS::HardDisk.$.File" ne 'ADFS::HardDisk.$.File'> and that
1294 the second stage of C<$> interpolation in regular expressions will fall
1295 foul of the C<$.> if scripts are not careful.
1297 Logical paths specified by system variables containing comma-separated
1298 search lists are also allowed; hence C<System:Modules> is a valid
1299 filename, and the filesystem will prefix C<Modules> with each section of
1300 C<System$Path> until a name is made that points to an object on disk.
1301 Writing to a new file C<System:Modules> would be allowed only if
1302 C<System$Path> contains a single item list. The filesystem will also
1303 expand system variables in filenames if enclosed in angle brackets, so
1304 C<< <System$Dir>.Modules >> would look for the file
1305 S<C<$ENV{'System$Dir'} . 'Modules'>>. The obvious implication of this is
1306 that B<fully qualified filenames can start with C<< <> >>> and should
1307 be protected when C<open> is used for input.
1309 Because C<.> was in use as a directory separator and filenames could not
1310 be assumed to be unique after 10 characters, Acorn implemented the C
1311 compiler to strip the trailing C<.c> C<.h> C<.s> and C<.o> suffix from
1312 filenames specified in source code and store the respective files in
1313 subdirectories named after the suffix. Hence files are translated:
1316 C:foo.h C:h.foo (logical path variable)
1317 sys/os.h sys.h.os (C compiler groks Unix-speak)
1318 10charname.c c.10charname
1319 10charname.o o.10charname
1320 11charname_.c c.11charname (assuming filesystem truncates at 10)
1322 The Unix emulation library's translation of filenames to native assumes
1323 that this sort of translation is required, and it allows a user-defined list
1324 of known suffixes that it will transpose in this fashion. This may
1325 seem transparent, but consider that with these rules F<foo/bar/baz.h>
1326 and F<foo/bar/h/baz> both map to F<foo.bar.h.baz>, and that C<readdir> and
1327 C<glob> cannot and do not attempt to emulate the reverse mapping. Other
1328 C<.>'s in filenames are translated to C</>.
1330 As implied above, the environment accessed through C<%ENV> is global, and
1331 the convention is that program specific environment variables are of the
1332 form C<Program$Name>. Each filesystem maintains a current directory,
1333 and the current filesystem's current directory is the B<global> current
1334 directory. Consequently, sociable programs don't change the current
1335 directory but rely on full pathnames, and programs (and Makefiles) cannot
1336 assume that they can spawn a child process which can change the current
1337 directory without affecting its parent (and everyone else for that
1340 Because native operating system filehandles are global and are currently
1341 allocated down from 255, with 0 being a reserved value, the Unix emulation
1342 library emulates Unix filehandles. Consequently, you can't rely on
1343 passing C<STDIN>, C<STDOUT>, or C<STDERR> to your children.
1345 The desire of users to express filenames of the form
1346 C<< <Foo$Dir>.Bar >> on the command line unquoted causes problems,
1347 too: C<``> command output capture has to perform a guessing game. It
1348 assumes that a string C<< <[^<>]+\$[^<>]> >> is a
1349 reference to an environment variable, whereas anything else involving
1350 C<< < >> or C<< > >> is redirection, and generally manages to be 99%
1351 right. Of course, the problem remains that scripts cannot rely on any
1352 Unix tools being available, or that any tools found have Unix-like command
1355 Extensions and XS are, in theory, buildable by anyone using free
1356 tools. In practice, many don't, as users of the Acorn platform are
1357 used to binary distributions. MakeMaker does run, but no available
1358 make currently copes with MakeMaker's makefiles; even if and when
1359 this should be fixed, the lack of a Unix-like shell will cause
1360 problems with makefile rules, especially lines of the form C<cd
1361 sdbm && make all>, and anything using quoting.
1363 "S<RISC OS>" is the proper name for the operating system, but the value
1364 in C<$^O> is "riscos" (because we don't like shouting).
1368 Perl has been ported to many platforms that do not fit into any of
1369 the categories listed above. Some, such as AmigaOS,
1370 QNX, Plan 9, and VOS, have been well-integrated into the standard
1371 Perl source code kit. You may need to see the F<ports/> directory
1372 on CPAN for information, and possibly binaries, for the likes of:
1373 aos, Atari ST, lynxos, riscos, Novell Netware, Tandem Guardian,
1374 I<etc.> (Yes, we know that some of these OSes may fall under the
1375 Unix category, but we are not a standards body.)
1377 Some approximate operating system names and their C<$^O> values
1378 in the "OTHER" category include:
1380 OS $^O $Config{'archname'}
1381 ------------------------------------------
1382 Amiga DOS amigaos m68k-amigos
1390 Amiga, F<README.amiga> (installed as L<perlamiga>).
1394 A free perl5-based PERL.NLM for Novell Netware is available in
1395 precompiled binary and source code form from L<http://www.novell.com/>
1396 as well as from CPAN.
1400 S<Plan 9>, F<README.plan9>
1404 =head1 FUNCTION IMPLEMENTATIONS
1406 Listed below are functions that are either completely unimplemented
1407 or else have been implemented differently on various platforms.
1408 Following each description will be, in parentheses, a list of
1409 platforms that the description applies to.
1411 The list may well be incomplete, or even wrong in some places. When
1412 in doubt, consult the platform-specific README files in the Perl
1413 source distribution, and any other documentation resources accompanying
1416 Be aware, moreover, that even among Unix-ish systems there are variations.
1418 For many functions, you can also query C<%Config>, exported by
1419 default from the C<Config> module. For example, to check whether the
1420 platform has the C<lstat> call, check C<$Config{d_lstat}>. See
1421 L<Config> for a full description of available variables.
1423 =head2 Alphabetical Listing of Perl Functions
1429 C<-w> only inspects the read-only file attribute (FILE_ATTRIBUTE_READONLY),
1430 which determines whether the directory can be deleted, not whether it can
1431 be written to. Directories always have read and write access unless denied
1432 by discretionary access control lists (DACLs). (S<Win32>)
1434 C<-r>, C<-w>, C<-x>, and C<-o> tell whether the file is accessible,
1435 which may not reflect UIC-based file protections. (VMS)
1437 C<-s> by name on an open file will return the space reserved on disk,
1438 rather than the current extent. C<-s> on an open filehandle returns the
1439 current size. (S<RISC OS>)
1441 C<-R>, C<-W>, C<-X>, C<-O> are indistinguishable from C<-r>, C<-w>,
1442 C<-x>, C<-o>. (Win32, VMS, S<RISC OS>)
1444 C<-g>, C<-k>, C<-l>, C<-u>, C<-A> are not particularly meaningful.
1445 (Win32, VMS, S<RISC OS>)
1447 C<-p> is not particularly meaningful. (VMS, S<RISC OS>)
1449 C<-d> is true if passed a device spec without an explicit directory.
1452 C<-x> (or C<-X>) determine if a file ends in one of the executable
1453 suffixes. C<-S> is meaningless. (Win32)
1455 C<-x> (or C<-X>) determine if a file has an executable file type.
1460 Emulated using timers that must be explicitly polled whenever Perl
1461 wants to dispatch "safe signals" and therefore cannot interrupt
1462 blocking system calls. (Win32)
1466 Due to issues with various CPUs, math libraries, compilers, and standards,
1467 results for C<atan2()> may vary depending on any combination of the above.
1468 Perl attempts to conform to the Open Group/IEEE standards for the results
1469 returned from C<atan2()>, but cannot force the issue if the system Perl is
1470 run on does not allow it. (Tru64, HP-UX 10.20)
1472 The current version of the standards for C<atan2()> is available at
1473 L<http://www.opengroup.org/onlinepubs/009695399/functions/atan2.html>.
1477 Meaningless. (S<RISC OS>)
1479 Reopens file and restores pointer; if function fails, underlying
1480 filehandle may be closed, or pointer may be in a different position.
1483 The value returned by C<tell> may be affected after the call, and
1484 the filehandle may be flushed. (Win32)
1488 Only good for changing "owner" read-write access, "group", and "other"
1489 bits are meaningless. (Win32)
1491 Only good for changing "owner" and "other" read-write access. (S<RISC OS>)
1493 Access permissions are mapped onto VOS access-control list changes. (VOS)
1495 The actual permissions set depend on the value of the C<CYGWIN>
1496 in the SYSTEM environment settings. (Cygwin)
1498 Setting the exec bit on some locations (generally F</sdcard>) will return true
1499 but not actually set the bit. (Android)
1503 Not implemented. (Win32, S<Plan 9>, S<RISC OS>)
1505 Does nothing, but won't fail. (Win32)
1507 A little funky, because VOS's notion of ownership is a little funky (VOS).
1511 Not implemented. (Win32, VMS, S<Plan 9>, S<RISC OS>, VOS)
1515 May not be available if library or source was not provided when building
1518 Not implemented. (Android)
1522 Not implemented. (VMS, S<Plan 9>, VOS)
1526 Not implemented. (VMS, S<Plan 9>, VOS)
1530 Not useful. (S<RISC OS>)
1532 Not supported. (Cygwin, Win32)
1534 Invokes VMS debugger. (VMS)
1538 C<exec LIST> without the use of indirect object syntax (C<exec PROGRAM LIST>)
1539 may fall back to trying the shell if the first C<spawn()> fails. (Win32)
1541 Does not automatically flush output handles on some platforms.
1542 (SunOS, Solaris, HP-UX)
1544 Not supported. (Symbian OS)
1548 Emulates Unix C<exit()> (which considers C<exit 1> to indicate an error) by
1549 mapping the C<1> to C<SS$_ABORT> (C<44>). This behavior may be overridden
1550 with the pragma C<use vmsish 'exit'>. As with the CRTL's C<exit()>
1551 function, C<exit 0> is also mapped to an exit status of C<SS$_NORMAL>
1552 (C<1>); this mapping cannot be overridden. Any other argument to
1554 is used directly as Perl's exit status. On VMS, unless the future
1555 POSIX_EXIT mode is enabled, the exit code should always be a valid
1556 VMS exit code and not a generic number. When the POSIX_EXIT mode is
1557 enabled, a generic number will be encoded in a method compatible with
1558 the C library _POSIX_EXIT macro so that it can be decoded by other
1559 programs, particularly ones written in C, like the GNV package. (VMS)
1561 C<exit()> resets file pointers, which is a problem when called
1562 from a child process (created by C<fork()>) in C<BEGIN>.
1563 A workaround is to use C<POSIX::_exit>. (Solaris)
1565 exit unless $Config{archname} =~ /\bsolaris\b/;
1566 require POSIX and POSIX::_exit(0);
1570 Not implemented. (Win32)
1572 Some functions available based on the version of VMS. (VMS)
1576 Not implemented (VMS, S<RISC OS>, VOS).
1580 Not implemented. (AmigaOS, S<RISC OS>, VMS)
1582 Emulated using multiple interpreters. See L<perlfork>. (Win32)
1584 Does not automatically flush output handles on some platforms.
1585 (SunOS, Solaris, HP-UX)
1589 Not implemented. (S<RISC OS>)
1593 Not implemented. (Win32, VMS, S<RISC OS>)
1597 Not implemented. (Win32, S<RISC OS>)
1601 Not implemented. (Win32, VMS, S<RISC OS>, VOS)
1605 Not implemented. (Win32)
1607 Not useful. (S<RISC OS>)
1611 Not implemented. (Win32, VMS, S<RISC OS>)
1615 Not implemented. (Android, Win32, S<Plan 9>)
1619 Not implemented. (Win32)
1621 Not useful. (S<RISC OS>)
1625 Not implemented. (Win32, VMS, S<RISC OS>)
1629 Not implemented. (Android, Win32, S<Plan 9>)
1631 =item getprotobynumber
1633 Not implemented. (Android)
1639 Not implemented. (Android, Win32)
1643 Not implemented. (Android, Win32, VMS)
1647 C<gethostbyname('localhost')> does not work everywhere: you may have
1648 to use C<gethostbyname('127.0.0.1')>. (S<Irix 5>)
1652 Not implemented. (Win32)
1656 Not implemented. (Android, Win32, S<Plan 9>)
1660 Not implemented. (Android, Win32, S<Plan 9>)
1664 Not implemented. (Win32, S<Plan 9>)
1668 Not implemented. (Android)
1672 Not implemented. (Android, Win32, S<Plan 9>, S<RISC OS>)
1676 Not implemented. (Win32, S<Plan 9>, S<RISC OS>)
1680 Not implemented. (Android, Win32, S<Plan 9>, S<RISC OS>)
1684 Not implemented. (S<Plan 9>, Win32, S<RISC OS>)
1688 Not implemented. (Win32)
1690 Either not implemented or a no-op. (Android)
1694 Not implemented. (Android, S<RISC OS>, VMS, Win32)
1698 Not implemented. (Android, Win32)
1702 Not implemented. (Android, Win32, S<Plan 9>)
1706 Not implemented. (Android, Win32, S<Plan 9>)
1710 Not implemented. (S<Plan 9>, Win32)
1712 =item getsockopt SOCKET,LEVEL,OPTNAME
1714 Not implemented. (S<Plan 9>)
1718 This operator is implemented via the C<File::Glob> extension on most
1719 platforms. See L<File::Glob> for portability information.
1723 In theory, C<gmtime()> is reliable from -2**63 to 2**63-1. However,
1724 because work arounds in the implementation use floating point numbers,
1725 it will become inaccurate as the time gets larger. This is a bug and
1726 will be fixed in the future.
1728 On VOS, time values are 32-bit quantities.
1730 =item ioctl FILEHANDLE,FUNCTION,SCALAR
1732 Not implemented. (VMS)
1734 Available only for socket handles, and it does what the C<ioctlsocket()> call
1735 in the Winsock API does. (Win32)
1737 Available only for socket handles. (S<RISC OS>)
1741 Not implemented, hence not useful for taint checking. (S<RISC OS>)
1743 C<kill()> doesn't have the semantics of C<raise()>, i.e. it doesn't send
1744 a signal to the identified process like it does on Unix platforms.
1745 Instead C<kill($sig, $pid)> terminates the process identified by C<$pid>,
1746 and makes it exit immediately with exit status $sig. As in Unix, if
1747 $sig is 0 and the specified process exists, it returns true without
1748 actually terminating it. (Win32)
1750 C<kill(-9, $pid)> will terminate the process specified by C<$pid> and
1751 recursively all child processes owned by it. This is different from
1752 the Unix semantics, where the signal will be delivered to all
1753 processes in the same process group as the process specified by
1756 Is not supported for process identification number of 0 or negative
1761 Not implemented. (S<RISC OS>, VOS)
1763 Link count not updated because hard links are not quite that hard
1764 (They are sort of half-way between hard and soft links). (AmigaOS)
1766 Hard links are implemented on Win32 under NTFS only. They are
1767 natively supported on Windows 2000 and later. On Windows NT they
1768 are implemented using the Windows POSIX subsystem support and the
1769 Perl process will need Administrator or Backup Operator privileges
1770 to create hard links.
1772 Available on 64 bit OpenVMS 8.2 and later. (VMS)
1776 localtime() has the same range as L</gmtime>, but because time zone
1777 rules change its accuracy for historical and future times may degrade
1778 but usually by no more than an hour.
1782 Not implemented. (S<RISC OS>)
1784 Return values (especially for device and inode) may be bogus. (Win32)
1794 Not implemented. (Android, Win32, VMS, S<Plan 9>, S<RISC OS>, VOS)
1798 open to C<|-> and C<-|> are unsupported. (Win32, S<RISC OS>)
1800 Opening a process does not automatically flush output handles on some
1801 platforms. (SunOS, Solaris, HP-UX)
1805 Not implemented. (Win32, VMS, S<RISC OS>)
1809 Can't move directories between directories on different logical volumes. (Win32)
1813 Will not cause C<readdir()> to re-read the directory stream. The entries
1814 already read before the C<rewinddir()> call will just be returned again
1815 from a cache buffer. (Win32)
1819 Only implemented on sockets. (Win32, VMS)
1821 Only reliable on sockets. (S<RISC OS>)
1823 Note that the C<select FILEHANDLE> form is generally portable.
1831 Not implemented. (Android, Win32, VMS, S<RISC OS>)
1835 Not implemented. (Android, VMS, Win32, S<RISC OS>)
1839 Not implemented. (Win32, VMS, S<RISC OS>, VOS)
1843 Not implemented. (Win32, VMS, S<RISC OS>, VOS)
1847 Not implemented. (Android, Win32, S<RISC OS>)
1851 Not implemented. (S<Plan 9>)
1861 Not implemented. (Android, Win32, VMS, S<RISC OS>)
1865 Emulated using synchronization functions such that it can be
1866 interrupted by C<alarm()>, and limited to a maximum of 4294967 seconds,
1867 approximately 49 days. (Win32)
1871 A relatively recent addition to socket functions, may not
1872 be implemented even in Unix platforms.
1876 Not implemented. (S<RISC OS>)
1878 Available on 64 bit OpenVMS 8.2 and later. (VMS)
1882 Platforms that do not have rdev, blksize, or blocks will return these
1883 as '', so numeric comparison or manipulation of these fields may cause
1884 'not numeric' warnings.
1886 ctime not supported on UFS (S<Mac OS X>).
1888 ctime is creation time instead of inode change time (Win32).
1890 device and inode are not meaningful. (Win32)
1892 device and inode are not necessarily reliable. (VMS)
1894 mtime, atime and ctime all return the last modification time. Device and
1895 inode are not necessarily reliable. (S<RISC OS>)
1897 dev, rdev, blksize, and blocks are not available. inode is not
1898 meaningful and will differ between stat calls on the same file. (os2)
1900 some versions of cygwin when doing a C<stat("foo")> and if not finding it
1901 may then attempt to C<stat("foo.exe")> (Cygwin)
1903 On Win32 C<stat()> needs to open the file to determine the link count
1904 and update attributes that may have been changed through hard links.
1905 Setting C<${^WIN32_SLOPPY_STAT}> to a true value speeds up C<stat()> by
1906 not performing this operation. (Win32)
1910 Not implemented. (Win32, S<RISC OS>)
1912 Implemented on 64 bit VMS 8.3. VMS requires the symbolic link to be in Unix
1913 syntax if it is intended to resolve to a valid path.
1917 Not implemented. (Win32, VMS, S<RISC OS>, VOS)
1921 The traditional "0", "1", and "2" MODEs are implemented with different
1922 numeric values on some systems. The flags exported by C<Fcntl>
1923 (O_RDONLY, O_WRONLY, O_RDWR) should work everywhere though. (S<Mac
1928 As an optimization, may not call the command shell specified in
1929 C<$ENV{PERL5SHELL}>. C<system(1, @args)> spawns an external
1930 process and immediately returns its process designator, without
1931 waiting for it to terminate. Return value may be used subsequently
1932 in C<wait> or C<waitpid>. Failure to C<spawn()> a subprocess is indicated
1933 by setting C<$?> to S<C<"255 << 8">>. C<$?> is set in a way compatible with
1934 Unix (i.e. the exitstatus of the subprocess is obtained by S<C<"$? >> 8">>,
1935 as described in the documentation). (Win32)
1937 There is no shell to process metacharacters, and the native standard is
1938 to pass a command line terminated by "\n" "\r" or "\0" to the spawned
1939 program. Redirection such as C<< > foo >> is performed (if at all) by
1940 the run time library of the spawned program. C<system> I<list> will call
1941 the Unix emulation library's C<exec> emulation, which attempts to provide
1942 emulation of the stdin, stdout, stderr in force in the parent, providing
1943 the child program uses a compatible version of the emulation library.
1944 I<scalar> will call the native command line direct and no such emulation
1945 of a child Unix program will exists. Mileage B<will> vary. (S<RISC OS>)
1947 C<system LIST> without the use of indirect object syntax (C<system PROGRAM LIST>)
1948 may fall back to trying the shell if the first C<spawn()> fails. (Win32)
1950 Does not automatically flush output handles on some platforms.
1951 (SunOS, Solaris, HP-UX)
1953 The return value is POSIX-like (shifted up by 8 bits), which only allows
1954 room for a made-up value derived from the severity bits of the native
1955 32-bit condition code (unless overridden by C<use vmsish 'status'>).
1956 If the native condition code is one that has a POSIX value encoded, the
1957 POSIX value will be decoded to extract the expected exit value.
1958 For more details see L<perlvms/$?>. (VMS)
1962 Not implemented. (Android)
1966 "cumulative" times will be bogus. On anything other than Windows NT
1967 or Windows 2000, "system" time will be bogus, and "user" time is
1968 actually the time returned by the C<clock()> function in the C runtime
1971 Not useful. (S<RISC OS>)
1975 Not implemented. (Older versions of VMS)
1977 Truncation to same-or-shorter lengths only. (VOS)
1979 If a FILEHANDLE is supplied, it must be writable and opened in append
1980 mode (i.e., use C<<< open(FH, '>>filename') >>>
1981 or C<sysopen(FH,...,O_APPEND|O_RDWR)>. If a filename is supplied, it
1982 should not be held open elsewhere. (Win32)
1986 Returns undef where unavailable.
1988 C<umask> works but the correct permissions are set only when the file
1989 is finally closed. (AmigaOS)
1993 Only the modification time is updated. (VMS, S<RISC OS>)
1995 May not behave as expected. Behavior depends on the C runtime
1996 library's implementation of C<utime()>, and the filesystem being
1997 used. The FAT filesystem typically does not support an "access
1998 time" field, and it may limit timestamps to a granularity of
1999 two seconds. (Win32)
2005 Can only be applied to process handles returned for processes spawned
2006 using C<system(1, ...)> or pseudo processes created with C<fork()>. (Win32)
2008 Not useful. (S<RISC OS>)
2013 =head1 Supported Platforms
2015 The following platforms are known to build Perl 5.12 (as of April 2010,
2016 its release date) from the standard source code distribution available
2017 at L<http://www.cpan.org/src>
2021 =item Linux (x86, ARM, IA64)
2035 =item Windows Server 2003
2039 =item Windows Server 2008
2047 Some tests are known to fail:
2053 F<ext/XS-APItes/t/call_checker.t> - see
2054 L<https://rt.perl.org/Ticket/Display.html?id=78502>
2058 F<dist/I18N-Collate/t/I18N-Collate.t>
2062 F<ext/Win32CORE/t/win32core.t> - may fail on recent cygwin installs.
2066 =item Solaris (x86, SPARC)
2072 =item Alpha (7.2 and later)
2074 =item I64 (8.2 and later)
2084 =item Debian GNU/kFreeBSD
2088 =item Irix (6.5. What else?)
2096 =item QNX Neutrino RTOS (6.5.0)
2100 =item Stratus OpenVOS (17.0 or later)
2106 =item time_t issues that may or may not be fixed
2110 =item Symbian (Series 60 v3, 3.2 and 5 - what else?)
2112 =item Stratus VOS / OpenVOS
2120 Perl now builds with FreeMiNT/Atari. It fails a few tests, that needs
2123 The FreeMiNT port uses GNU dld for loadable module capabilities. So
2124 ensure you have that library installed when building perl.
2128 =head1 EOL Platforms
2132 The following platforms were supported by a previous version of
2133 Perl but have been officially removed from Perl's source code
2144 The following platforms were supported up to 5.10. They may still
2145 have worked in 5.12, but supporting code has been removed for 5.14:
2161 The following platforms were supported by a previous version of
2162 Perl but have been officially removed from Perl's source code
2169 =item Apollo Domain/OS
2171 =item Apple Mac OS 8/9
2178 =head1 Supported Platforms (Perl 5.8)
2180 As of July 2002 (the Perl release 5.8.0), the following platforms were
2181 able to build Perl from the standard source code distribution
2182 available at L<http://www.cpan.org/src/>
2193 HI-UXMPP (Hitachi) (5.8.0 worked but we didn't know it)
2203 ReliantUNIX (formerly SINIX)
2205 OpenVMS (formerly VMS)
2206 Open UNIX (Unixware) (since Perl 5.8.1/5.9.0)
2208 OS/400 (using the PASE) (since Perl 5.8.1/5.9.0)
2210 POSIX-BC (formerly BS2000)
2215 Tru64 UNIX (formerly DEC OSF/1, Digital UNIX)
2220 Win95/98/ME/2K/XP 2)
2222 z/OS (formerly OS/390)
2225 1) in DOS mode either the DOS or OS/2 ports can be used
2226 2) compilers: Borland, MinGW (GCC), VC6
2228 The following platforms worked with the previous releases (5.6 and
2229 5.7), but we did not manage either to fix or to test these in time
2230 for the 5.8.0 release. There is a very good chance that many of these
2231 will work fine with the 5.8.0.
2244 Known to be broken for 5.8.0 (but 5.6.1 and 5.7.2 can be used):
2248 The following platforms have been known to build Perl from source in
2249 the past (5.005_03 and earlier), but we haven't been able to verify
2250 their status for the current release, either because the
2251 hardware/software platforms are rare or because we don't have an
2252 active champion on these platforms--or both. They used to work,
2253 though, so go ahead and try compiling them, and let perlbug@perl.org
2286 The following platforms have their own source code distributions and
2287 binaries available via L<http://www.cpan.org/ports/>
2291 OS/400 (ILE) 5.005_02
2292 Tandem Guardian 5.004
2294 The following platforms have only binaries available via
2295 L<http://www.cpan.org/ports/index.html> :
2299 Acorn RISCOS 5.005_02
2303 Although we do suggest that you always build your own Perl from
2304 the source code, both for maximal configurability and for security,
2305 in case you are in a hurry you can check
2306 L<http://www.cpan.org/ports/index.html> for binary distributions.
2310 L<perlaix>, L<perlamiga>, L<perlbs2000>,
2311 L<perlce>, L<perlcygwin>, L<perldos>,
2312 L<perlebcdic>, L<perlfreebsd>, L<perlhurd>, L<perlhpux>, L<perlirix>,
2313 L<perlmacos>, L<perlmacosx>,
2314 L<perlnetware>, L<perlos2>, L<perlos390>, L<perlos400>,
2315 L<perlplan9>, L<perlqnx>, L<perlsolaris>, L<perltru64>,
2316 L<perlunicode>, L<perlvms>, L<perlvos>, L<perlwin32>, and L<Win32>.
2318 =head1 AUTHORS / CONTRIBUTORS
2320 Abigail <abigail@foad.org>,
2321 Charles Bailey <bailey@newman.upenn.edu>,
2322 Graham Barr <gbarr@pobox.com>,
2323 Tom Christiansen <tchrist@perl.com>,
2324 Nicholas Clark <nick@ccl4.org>,
2325 Thomas Dorner <Thomas.Dorner@start.de>,
2326 Andy Dougherty <doughera@lafayette.edu>,
2327 Dominic Dunlop <domo@computer.org>,
2328 Neale Ferguson <neale@vma.tabnsw.com.au>,
2329 David J. Fiander <davidf@mks.com>,
2330 Paul Green <Paul.Green@stratus.com>,
2331 M.J.T. Guy <mjtg@cam.ac.uk>,
2332 Jarkko Hietaniemi <jhi@iki.fi>,
2333 Luther Huffman <lutherh@stratcom.com>,
2334 Nick Ing-Simmons <nick@ing-simmons.net>,
2335 Andreas J. KE<ouml>nig <a.koenig@mind.de>,
2336 Markus Laker <mlaker@contax.co.uk>,
2337 Andrew M. Langmead <aml@world.std.com>,
2338 Larry Moore <ljmoore@freespace.net>,
2339 Paul Moore <Paul.Moore@uk.origin-it.com>,
2340 Chris Nandor <pudge@pobox.com>,
2341 Matthias Neeracher <neeracher@mac.com>,
2342 Philip Newton <pne@cpan.org>,
2343 Gary Ng <71564.1743@CompuServe.COM>,
2344 Tom Phoenix <rootbeer@teleport.com>,
2345 AndrE<eacute> Pirard <A.Pirard@ulg.ac.be>,
2346 Peter Prymmer <pvhp@forte.com>,
2347 Hugo van der Sanden <hv@crypt0.demon.co.uk>,
2348 Gurusamy Sarathy <gsar@activestate.com>,
2349 Paul J. Schinder <schinder@pobox.com>,
2350 Michael G Schwern <schwern@pobox.com>,
2351 Dan Sugalski <dan@sidhe.org>,
2352 Nathan Torkington <gnat@frii.com>,
2353 John Malmberg <wb8tyw@qsl.net>