3 perlport - Writing portable Perl
7 Perl runs on numerous operating systems. While most of them share
8 much in common, they also have their own unique features.
10 This document is meant to help you to find out what constitutes portable
11 Perl code. That way once you make a decision to write portably,
12 you know where the lines are drawn, and you can stay within them.
14 There is a tradeoff between taking full advantage of one particular
15 type of computer and taking advantage of a full range of them.
16 Naturally, as you broaden your range and become more diverse, the
17 common factors drop, and you are left with an increasingly smaller
18 area of common ground in which you can operate to accomplish a
19 particular task. Thus, when you begin attacking a problem, it is
20 important to consider under which part of the tradeoff curve you
21 want to operate. Specifically, you must decide whether it is
22 important that the task that you are coding have the full generality
23 of being portable, or whether to just get the job done right now.
24 This is the hardest choice to be made. The rest is easy, because
25 Perl provides many choices, whichever way you want to approach your
28 Looking at it another way, writing portable code is usually about
29 willfully limiting your available choices. Naturally, it takes
30 discipline and sacrifice to do that. The product of portability
31 and convenience may be a constant. You have been warned.
33 Be aware of two important points:
37 =item Not all Perl programs have to be portable
39 There is no reason you should not use Perl as a language to glue Unix
40 tools together, or to prototype a Macintosh application, or to manage the
41 Windows registry. If it makes no sense to aim for portability for one
42 reason or another in a given program, then don't bother.
44 =item Nearly all of Perl already I<is> portable
46 Don't be fooled into thinking that it is hard to create portable Perl
47 code. It isn't. Perl tries its level-best to bridge the gaps between
48 what's available on different platforms, and all the means available to
49 use those features. Thus almost all Perl code runs on any machine
50 without modification. But there are some significant issues in
51 writing portable code, and this document is entirely about those issues.
55 Here's the general rule: When you approach a task commonly done
56 using a whole range of platforms, think about writing portable
57 code. That way, you don't sacrifice much by way of the implementation
58 choices you can avail yourself of, and at the same time you can give
59 your users lots of platform choices. On the other hand, when you have to
60 take advantage of some unique feature of a particular platform, as is
61 often the case with systems programming (whether for Unix, Windows,
62 VMS, etc.), consider writing platform-specific code.
64 When the code will run on only two or three operating systems, you
65 may need to consider only the differences of those particular systems.
66 The important thing is to decide where the code will run and to be
67 deliberate in your decision.
69 The material below is separated into three main sections: main issues of
70 portability (L<"ISSUES">), platform-specific issues (L<"PLATFORMS">), and
71 built-in perl functions that behave differently on various ports
72 (L<"FUNCTION IMPLEMENTATIONS">).
74 This information should not be considered complete; it includes possibly
75 transient information about idiosyncrasies of some of the ports, almost
76 all of which are in a state of constant evolution. Thus, this material
77 should be considered a perpetual work in progress
78 (C<< <IMG SRC="yellow_sign.gif" ALT="Under Construction"> >>).
84 In most operating systems, lines in files are terminated by newlines.
85 Just what is used as a newline may vary from OS to OS. Unix
86 traditionally uses C<\012>, one type of DOSish I/O uses C<\015\012>,
87 and S<Mac OS> uses C<\015>.
89 Perl uses C<\n> to represent the "logical" newline, where what is
90 logical may depend on the platform in use. In MacPerl, C<\n> always
91 means C<\015>. 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 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 $/ to the appropriate value for your file format
104 before using chomp().
106 Because of the "text" mode translation, DOSish perls have limitations
107 in using C<seek> and C<tell> on a file accessed in "text" mode.
108 Stick to C<seek>-ing to locations you got from C<tell> (and no
109 others), and you are usually free to use C<seek> and C<tell> even
110 in "text" mode. Using C<seek> or C<tell> or other file operations
111 may be non-portable. If you use C<binmode> on a file, however, you
112 can usually C<seek> and C<tell> with arbitrary values in safety.
114 A common misconception in socket programming is that C<\n> eq C<\012>
115 everywhere. When using protocols such as common Internet protocols,
116 C<\012> and C<\015> are called for specifically, and the values of
117 the logical C<\n> and C<\r> (carriage return) are not reliable.
119 print SOCKET "Hi there, client!\r\n"; # WRONG
120 print SOCKET "Hi there, client!\015\012"; # RIGHT
122 However, using C<\015\012> (or C<\cM\cJ>, or C<\x0D\x0A>) can be tedious
123 and unsightly, as well as confusing to those maintaining the code. As
124 such, the Socket module supplies the Right Thing for those who want it.
126 use Socket qw(:DEFAULT :crlf);
127 print SOCKET "Hi there, client!$CRLF" # RIGHT
129 When reading from a socket, remember that the default input record
130 separator C<$/> is C<\n>, but robust socket code will recognize as
131 either C<\012> or C<\015\012> as end of line:
137 Because both CRLF and LF end in LF, the input record separator can
138 be set to LF and any CR stripped later. Better to write:
140 use Socket qw(:DEFAULT :crlf);
141 local($/) = 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 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 Data::Dumper and Storable
241 (included as of perl 5.8). Keeping all data as text significantly
244 The v-strings are portable only up to v2147483647 (0x7FFFFFFF), that's
245 how far EBCDIC, or more precisely UTF-EBCDIC will go.
247 =head2 Files and Filesystems
249 Most platforms these days structure files in a hierarchical fashion.
250 So, it is reasonably safe to assume that all platforms support the
251 notion of a "path" to uniquely identify a file on the system. How
252 that path is really written, though, differs considerably.
254 Although similar, file path specifications differ between Unix,
255 Windows, S<Mac OS>, OS/2, VMS, VOS, S<RISC OS>, and probably others.
256 Unix, for example, is one of the few OSes that has the elegant idea
257 of a single root directory.
259 DOS, OS/2, VMS, VOS, and Windows can work similarly to Unix with C</>
260 as path separator, or in their own idiosyncratic ways (such as having
261 several root directories and various "unrooted" device files such NIL:
264 S<Mac OS> 9 and earlier used C<:> as a path separator instead of C</>.
266 The filesystem may support neither hard links (C<link>) nor
267 symbolic links (C<symlink>, C<readlink>, C<lstat>).
269 The filesystem may support neither access timestamp nor change
270 timestamp (meaning that about the only portable timestamp is the
271 modification timestamp), or one second granularity of any timestamps
272 (e.g. the FAT filesystem limits the time granularity to two seconds).
274 The "inode change timestamp" (the C<-C> filetest) may really be the
275 "creation timestamp" (which it is not in Unix).
277 VOS perl can emulate Unix filenames with C</> as path separator. The
278 native pathname characters greater-than, less-than, number-sign, and
279 percent-sign are always accepted.
281 S<RISC OS> perl can emulate Unix filenames with C</> as path
282 separator, or go native and use C<.> for path separator and C<:> to
283 signal filesystems and disk names.
285 Don't assume Unix filesystem access semantics: that read, write,
286 and execute are all the permissions there are, and even if they exist,
287 that their semantics (for example what do r, w, and x mean on
288 a directory) are the Unix ones. The various Unix/POSIX compatibility
289 layers usually try to make interfaces like chmod() work, but sometimes
290 there simply is no good mapping.
292 If all this is intimidating, have no (well, maybe only a little)
293 fear. There are modules that can help. The File::Spec modules
294 provide methods to do the Right Thing on whatever platform happens
295 to be running the program.
297 use File::Spec::Functions;
298 chdir(updir()); # go up one directory
299 my $file = catfile(curdir(), 'temp', 'file.txt');
300 # on Unix and Win32, './temp/file.txt'
301 # on Mac OS Classic, ':temp:file.txt'
302 # on VMS, '[.temp]file.txt'
304 File::Spec is available in the standard distribution as of version
305 5.004_05. File::Spec::Functions is only in File::Spec 0.7 and later,
306 and some versions of perl come with version 0.6. If File::Spec
307 is not updated to 0.7 or later, you must use the object-oriented
308 interface from File::Spec (or upgrade File::Spec).
310 In general, production code should not have file paths hardcoded.
311 Making them user-supplied or read from a configuration file is
312 better, keeping in mind that file path syntax varies on different
315 This is especially noticeable in scripts like Makefiles and test suites,
316 which often assume C</> as a path separator for subdirectories.
318 Also of use is File::Basename from the standard distribution, which
319 splits a pathname into pieces (base filename, full path to directory,
322 Even when on a single platform (if you can call Unix a single platform),
323 remember not to count on the existence or the contents of particular
324 system-specific files or directories, like F</etc/passwd>,
325 F</etc/sendmail.conf>, F</etc/resolv.conf>, or even F</tmp/>. For
326 example, F</etc/passwd> may exist but not contain the encrypted
327 passwords, because the system is using some form of enhanced security.
328 Or it may not contain all the accounts, because the system is using NIS.
329 If code does need to rely on such a file, include a description of the
330 file and its format in the code's documentation, then make it easy for
331 the user to override the default location of the file.
333 Don't assume a text file will end with a newline. They should,
336 Do not have two files or directories of the same name with different
337 case, like F<test.pl> and F<Test.pl>, as many platforms have
338 case-insensitive (or at least case-forgiving) filenames. Also, try
339 not to have non-word characters (except for C<.>) in the names, and
340 keep them to the 8.3 convention, for maximum portability, onerous a
341 burden though this may appear.
343 Likewise, when using the AutoSplit module, try to keep your functions to
344 8.3 naming and case-insensitive conventions; or, at the least,
345 make it so the resulting files have a unique (case-insensitively)
348 Whitespace in filenames is tolerated on most systems, but not all,
349 and even on systems where it might be tolerated, some utilities
350 might become confused by such whitespace.
352 Many systems (DOS, VMS ODS-2) cannot have more than one C<.> in their
355 Don't assume C<< > >> won't be the first character of a filename.
356 Always use C<< < >> explicitly to open a file for reading, or even
357 better, use the three-arg version of open, unless you want the user to
358 be able to specify a pipe open.
360 open my $fh, '<', $existing_file) or die $!;
362 If filenames might use strange characters, it is safest to open it
363 with C<sysopen> instead of C<open>. C<open> is magic and can
364 translate characters like C<< > >>, C<< < >>, and C<|>, which may
365 be the wrong thing to do. (Sometimes, though, it's the right thing.)
366 Three-arg open can also help protect against this translation in cases
367 where it is undesirable.
369 Don't use C<:> as a part of a filename since many systems use that for
370 their own semantics (Mac OS Classic for separating pathname components,
371 many networking schemes and utilities for separating the nodename and
372 the pathname, and so on). For the same reasons, avoid C<@>, C<;> and
375 Don't assume that in pathnames you can collapse two leading slashes
376 C<//> into one: some networking and clustering filesystems have special
377 semantics for that. Let the operating system to sort it out.
379 The I<portable filename characters> as defined by ANSI C are
381 a b c d e f g h i j k l m n o p q r t u v w x y z
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
386 and the "-" shouldn't be the first character. If you want to be
387 hypercorrect, stay case-insensitive and within the 8.3 naming
388 convention (all the files and directories have to be unique within one
389 directory if their names are lowercased and truncated to eight
390 characters before the C<.>, if any, and to three characters after the
391 C<.>, if any). (And do not use C<.>s in directory names.)
393 =head2 System Interaction
395 Not all platforms provide a command line. These are usually platforms
396 that rely primarily on a Graphical User Interface (GUI) for user
397 interaction. A program requiring a command line interface might
398 not work everywhere. This is probably for the user of the program
399 to deal with, so don't stay up late worrying about it.
401 Some platforms can't delete or rename files held open by the system,
402 this limitation may also apply to changing filesystem metainformation
403 like file permissions or owners. Remember to C<close> files when you
404 are done with them. Don't C<unlink> or C<rename> an open file. Don't
405 C<tie> or C<open> a file already tied or opened; C<untie> or C<close>
408 Don't open the same file more than once at a time for writing, as some
409 operating systems put mandatory locks on such files.
411 Don't assume that write/modify permission on a directory gives the
412 right to add or delete files/directories in that directory. That is
413 filesystem specific: in some filesystems you need write/modify
414 permission also (or even just) in the file/directory itself. In some
415 filesystems (AFS, DFS) the permission to add/delete directory entries
416 is a completely separate permission.
418 Don't assume that a single C<unlink> completely gets rid of the file:
419 some filesystems (most notably the ones in VMS) have versioned
420 filesystems, and unlink() removes only the most recent one (it doesn't
421 remove all the versions because by default the native tools on those
422 platforms remove just the most recent version, too). The portable
423 idiom to remove all the versions of a file is
425 1 while unlink "file";
427 This will terminate if the file is undeleteable for some reason
428 (protected, not there, and so on).
430 Don't count on a specific environment variable existing in C<%ENV>.
431 Don't count on C<%ENV> entries being case-sensitive, or even
432 case-preserving. Don't try to clear %ENV by saying C<%ENV = ();>, or,
433 if you really have to, make it conditional on C<$^O ne 'VMS'> since in
434 VMS the C<%ENV> table is much more than a per-process key-value string
437 On VMS, some entries in the %ENV hash are dynamically created when
438 their key is used on a read if they did not previously exist. The
439 values for C<$ENV{HOME}>, C<$ENV{TERM}>, C<$ENV{HOME}>, and C<$ENV{USER}>,
440 are known to be dynamically generated. The specific names that are
441 dynamically generated may vary with the version of the C library on VMS,
442 and more may exist than is documented.
444 On VMS by default, changes to the %ENV hash are persistent after the process
445 exits. This can cause unintended issues.
447 Don't count on signals or C<%SIG> for anything.
449 Don't count on filename globbing. Use C<opendir>, C<readdir>, and
452 Don't count on per-program environment variables, or per-program current
455 Don't count on specific values of C<$!>, neither numeric nor
456 especially the strings values. Users may switch their locales causing
457 error messages to be translated into their languages. If you can
458 trust a POSIXish environment, you can portably use the symbols defined
459 by the Errno module, like ENOENT. And don't trust on the values of C<$!>
460 at all except immediately after a failed system call.
462 =head2 Command names versus file pathnames
464 Don't assume that the name used to invoke a command or program with
465 C<system> or C<exec> can also be used to test for the existence of the
466 file that holds the executable code for that command or program.
467 First, many systems have "internal" commands that are built-in to the
468 shell or OS and while these commands can be invoked, there is no
469 corresponding file. Second, some operating systems (e.g., Cygwin,
470 DJGPP, OS/2, and VOS) have required suffixes for executable files;
471 these suffixes are generally permitted on the command name but are not
472 required. Thus, a command like "perl" might exist in a file named
473 "perl", "perl.exe", or "perl.pm", depending on the operating system.
474 The variable "_exe" in the Config module holds the executable suffix,
475 if any. Third, the VMS port carefully sets up $^X and
476 $Config{perlpath} so that no further processing is required. This is
477 just as well, because the matching regular expression used below would
478 then have to deal with a possible trailing version number in the VMS
481 To convert $^X to a file pathname, taking account of the requirements
482 of the various operating system possibilities, say:
487 {$thisperl .= $Config{_exe} unless $thisperl =~ m/$Config{_exe}$/i;}
489 To convert $Config{perlpath} to a file pathname, say:
492 my $thisperl = $Config{perlpath};
494 {$thisperl .= $Config{_exe} unless $thisperl =~ m/$Config{_exe}$/i;}
498 Don't assume that you can reach the public Internet.
500 Don't assume that there is only one way to get through firewalls
501 to the public Internet.
503 Don't assume that you can reach outside world through any other port
504 than 80, or some web proxy. ftp is blocked by many firewalls.
506 Don't assume that you can send email by connecting to the local SMTP port.
508 Don't assume that you can reach yourself or any node by the name
509 'localhost'. The same goes for '127.0.0.1'. You will have to try both.
511 Don't assume that the host has only one network card, or that it
512 can't bind to many virtual IP addresses.
514 Don't assume a particular network device name.
516 Don't assume a particular set of ioctl()s will work.
518 Don't assume that you can ping hosts and get replies.
520 Don't assume that any particular port (service) will respond.
522 Don't assume that Sys::Hostname (or any other API or command) returns
523 either a fully qualified hostname or a non-qualified hostname: it all
524 depends on how the system had been configured. Also remember that for
525 things such as DHCP and NAT, the hostname you get back might not be
528 All the above "don't":s may look daunting, and they are, but the key
529 is to degrade gracefully if one cannot reach the particular network
530 service one wants. Croaking or hanging do not look very professional.
532 =head2 Interprocess Communication (IPC)
534 In general, don't directly access the system in code meant to be
535 portable. That means, no C<system>, C<exec>, C<fork>, C<pipe>,
536 C<``>, C<qx//>, C<open> with a C<|>, nor any of the other things
537 that makes being a perl hacker worth being.
539 Commands that launch external processes are generally supported on
540 most platforms (though many of them do not support any type of
541 forking). The problem with using them arises from what you invoke
542 them on. External tools are often named differently on different
543 platforms, may not be available in the same location, might accept
544 different arguments, can behave differently, and often present their
545 results in a platform-dependent way. Thus, you should seldom depend
546 on them to produce consistent results. (Then again, if you're calling
547 I<netstat -a>, you probably don't expect it to run on both Unix and CP/M.)
549 One especially common bit of Perl code is opening a pipe to B<sendmail>:
551 open(MAIL, '|/usr/lib/sendmail -t')
552 or die "cannot fork sendmail: $!";
554 This is fine for systems programming when sendmail is known to be
555 available. But it is not fine for many non-Unix systems, and even
556 some Unix systems that may not have sendmail installed. If a portable
557 solution is needed, see the various distributions on CPAN that deal
558 with it. Mail::Mailer and Mail::Send in the MailTools distribution are
559 commonly used, and provide several mailing methods, including mail,
560 sendmail, and direct SMTP (via Net::SMTP) if a mail transfer agent is
561 not available. Mail::Sendmail is a standalone module that provides
562 simple, platform-independent mailing.
564 The Unix System V IPC (C<msg*(), sem*(), shm*()>) is not available
565 even on all Unix platforms.
567 Do not use either the bare result of C<pack("N", 10, 20, 30, 40)> or
568 bare v-strings (such as C<v10.20.30.40>) to represent IPv4 addresses:
569 both forms just pack the four bytes into network order. That this
570 would be equal to the C language C<in_addr> struct (which is what the
571 socket code internally uses) is not guaranteed. To be portable use
572 the routines of the Socket extension, such as C<inet_aton()>,
573 C<inet_ntoa()>, and C<sockaddr_in()>.
575 The rule of thumb for portable code is: Do it all in portable Perl, or
576 use a module (that may internally implement it with platform-specific
577 code, but expose a common interface).
579 =head2 External Subroutines (XS)
581 XS code can usually be made to work with any platform, but dependent
582 libraries, header files, etc., might not be readily available or
583 portable, or the XS code itself might be platform-specific, just as Perl
584 code might be. If the libraries and headers are portable, then it is
585 normally reasonable to make sure the XS code is portable, too.
587 A different type of portability issue arises when writing XS code:
588 availability of a C compiler on the end-user's system. C brings
589 with it its own portability issues, and writing XS code will expose
590 you to some of those. Writing purely in Perl is an easier way to
593 =head2 Standard Modules
595 In general, the standard modules work across platforms. Notable
596 exceptions are the CPAN module (which currently makes connections to external
597 programs that may not be available), platform-specific modules (like
598 ExtUtils::MM_VMS), and DBM modules.
600 There is no one DBM module available on all platforms.
601 SDBM_File and the others are generally available on all Unix and DOSish
602 ports, but not in MacPerl, where only NBDM_File and DB_File are
605 The good news is that at least some DBM module should be available, and
606 AnyDBM_File will use whichever module it can find. Of course, then
607 the code needs to be fairly strict, dropping to the greatest common
608 factor (e.g., not exceeding 1K for each record), so that it will
609 work with any DBM module. See L<AnyDBM_File> for more details.
613 The system's notion of time of day and calendar date is controlled in
614 widely different ways. Don't assume the timezone is stored in C<$ENV{TZ}>,
615 and even if it is, don't assume that you can control the timezone through
616 that variable. Don't assume anything about the three-letter timezone
617 abbreviations (for example that MST would be the Mountain Standard Time,
618 it's been known to stand for Moscow Standard Time). If you need to
619 use timezones, express them in some unambiguous format like the
620 exact number of minutes offset from UTC, or the POSIX timezone
623 Don't assume that the epoch starts at 00:00:00, January 1, 1970,
624 because that is OS- and implementation-specific. It is better to
625 store a date in an unambiguous representation. The ISO 8601 standard
626 defines YYYY-MM-DD as the date format, or YYYY-MM-DDTHH:MM:SS
627 (that's a literal "T" separating the date from the time).
628 Please do use the ISO 8601 instead of making us guess what
629 date 02/03/04 might be. ISO 8601 even sorts nicely as-is.
630 A text representation (like "1987-12-18") can be easily converted
631 into an OS-specific value using a module like Date::Parse.
632 An array of values, such as those returned by C<localtime>, can be
633 converted to an OS-specific representation using Time::Local.
635 When calculating specific times, such as for tests in time or date modules,
636 it may be appropriate to calculate an offset for the epoch.
639 my $offset = Time::Local::timegm(0, 0, 0, 1, 0, 70);
641 The value for C<$offset> in Unix will be C<0>, but in Mac OS Classic
642 will be some large number. C<$offset> can then be added to a Unix time
643 value to get what should be the proper value on any system.
645 =head2 Character sets and character encoding
647 Assume very little about character sets.
649 Assume nothing about numerical values (C<ord>, C<chr>) of characters.
650 Do not use explicit code point ranges (like C<\xHH-\xHH)>. However,
651 starting in Perl v5.22, regular expression pattern bracketed character
652 class ranges specified like C<qr/[\N{U+HH}-\N{U+HH}]/> are portable.
653 You can portably use, for example, symbolic character classes like
656 Do not assume that the alphabetic characters are encoded contiguously
657 (in the numeric sense). There may be gaps. Special coding in Perl,
658 however, guarantees that all subsets of C<qr/[A-Z]/>, C<qr/[a-z]/>, and
659 C<qr/[0-9]/> behave as expected.
661 Do not assume anything about the ordering of the characters.
662 The lowercase letters may come before or after the uppercase letters;
663 the lowercase and uppercase may be interlaced so that both "a" and "A"
664 come before "b"; the accented and other international characters may
665 be interlaced so that E<auml> comes before "b".
667 =head2 Internationalisation
669 If you may assume POSIX (a rather large assumption), you may read
670 more about the POSIX locale system from L<perllocale>. The locale
671 system at least attempts to make things a little bit more portable,
672 or at least more convenient and native-friendly for non-English
673 users. The system affects character sets and encoding, and date
674 and time formatting--amongst other things.
676 If you really want to be international, you should consider Unicode.
677 See L<perluniintro> and L<perlunicode> for more information.
679 If you want to use non-ASCII bytes (outside the bytes 0x00..0x7f) in
680 the "source code" of your code, to be portable you have to be explicit
681 about what bytes they are. Someone might for example be using your
682 code under a UTF-8 locale, in which case random native bytes might be
683 illegal ("Malformed UTF-8 ...") This means that for example embedding
684 ISO 8859-1 bytes beyond 0x7f into your strings might cause trouble
685 later. If the bytes are native 8-bit bytes, you can use the C<bytes>
686 pragma. If the bytes are in a string (regular expressions being
687 curious strings), you can often also use the C<\xHH> or more portably,
688 the C<\N{U+HH}> notations instead
689 of embedding the bytes as-is. If you want to write your code in UTF-8,
692 =head2 System Resources
694 If your code is destined for systems with severely constrained (or
695 missing!) virtual memory systems then you want to be I<especially> mindful
696 of avoiding wasteful constructs such as:
698 my @lines = <$very_large_file>; # bad
700 while (<$fh>) {$file .= $_} # sometimes bad
701 my $file = join('', <$fh>); # better
703 The last two constructs may appear unintuitive to most people. The
704 first repeatedly grows a string, whereas the second allocates a
705 large chunk of memory in one go. On some systems, the second is
706 more efficient that the first.
710 Most multi-user platforms provide basic levels of security, usually
711 implemented at the filesystem level. Some, however, unfortunately do
712 not. Thus the notion of user id, or "home" directory,
713 or even the state of being logged-in, may be unrecognizable on many
714 platforms. If you write programs that are security-conscious, it
715 is usually best to know what type of system you will be running
716 under so that you can write code explicitly for that platform (or
719 Don't assume the Unix filesystem access semantics: the operating
720 system or the filesystem may be using some ACL systems, which are
721 richer languages than the usual rwx. Even if the rwx exist,
722 their semantics might be different.
724 (From security viewpoint testing for permissions before attempting to
725 do something is silly anyway: if one tries this, there is potential
726 for race conditions. Someone or something might change the
727 permissions between the permissions check and the actual operation.
728 Just try the operation.)
730 Don't assume the Unix user and group semantics: especially, don't
731 expect the C<< $< >> and C<< $> >> (or the C<$(> and C<$)>) to work
732 for switching identities (or memberships).
734 Don't assume set-uid and set-gid semantics. (And even if you do,
735 think twice: set-uid and set-gid are a known can of security worms.)
739 For those times when it is necessary to have platform-specific code,
740 consider keeping the platform-specific code in one place, making porting
741 to other platforms easier. Use the Config module and the special
742 variable C<$^O> to differentiate platforms, as described in
745 Be careful in the tests you supply with your module or programs.
746 Module code may be fully portable, but its tests might not be. This
747 often happens when tests spawn off other processes or call external
748 programs to aid in the testing, or when (as noted above) the tests
749 assume certain things about the filesystem and paths. Be careful not
750 to depend on a specific output style for errors, such as when checking
751 C<$!> after a failed system call. Using C<$!> for anything else than
752 displaying it as output is doubtful (though see the Errno module for
753 testing reasonably portably for error value). Some platforms expect
754 a certain output format, and Perl on those platforms may have been
755 adjusted accordingly. Most specifically, don't anchor a regex when
756 testing an error value.
760 Modules uploaded to CPAN are tested by a variety of volunteers on
761 different platforms. These CPAN testers are notified by mail of each
762 new upload, and reply to the list with PASS, FAIL, NA (not applicable to
763 this platform), or UNKNOWN (unknown), along with any relevant notations.
765 The purpose of the testing is twofold: one, to help developers fix any
766 problems in their code that crop up because of lack of testing on other
767 platforms; two, to provide users with information about whether
768 a given module works on a given platform.
776 Mailing list: cpan-testers-discuss@perl.org
780 Testing results: L<http://www.cpantesters.org/>
786 Perl is built with a C<$^O> variable that indicates the operating
787 system it was built on. This was implemented
788 to help speed up code that would otherwise have to C<use Config>
789 and use the value of C<$Config{osname}>. Of course, to get more
790 detailed information about the system, looking into C<%Config> is
791 certainly recommended.
793 C<%Config> cannot always be trusted, however, because it was built
794 at compile time. If perl was built in one place, then transferred
795 elsewhere, some values may be wrong. The values may even have been
796 edited after the fact.
800 Perl works on a bewildering variety of Unix and Unix-like platforms (see
801 e.g. most of the files in the F<hints/> directory in the source code kit).
802 On most of these systems, the value of C<$^O> (hence C<$Config{'osname'}>,
803 too) is determined either by lowercasing and stripping punctuation from the
804 first field of the string returned by typing C<uname -a> (or a similar command)
805 at the shell prompt or by testing the file system for the presence of
806 uniquely named files such as a kernel or header file. Here, for example,
807 are a few of the more popular Unix flavors:
809 uname $^O $Config{'archname'}
810 --------------------------------------------
812 BSD/OS bsdos i386-bsdos
814 DYNIX/ptx dynixptx i386-dynixptx
815 FreeBSD freebsd freebsd-i386
816 Haiku haiku BePC-haiku
817 Linux linux arm-linux
818 Linux linux armv5tel-linux
819 Linux linux i386-linux
820 Linux linux i586-linux
821 Linux linux ppc-linux
822 HP-UX hpux PA-RISC1.1
824 Mac OS X darwin darwin
826 NeXT 4 next OPENSTEP-Mach
827 openbsd openbsd i386-openbsd
828 OSF1 dec_osf alpha-dec_osf
829 reliantunix-n svr4 RM400-svr4
830 SCO_SV sco_sv i386-sco_sv
831 SINIX-N svr4 RM400-svr4
832 sn4609 unicos CRAY_C90-unicos
833 sn6521 unicosmk t3e-unicosmk
834 sn9617 unicos CRAY_J90-unicos
835 SunOS solaris sun4-solaris
836 SunOS solaris i86pc-solaris
837 SunOS4 sunos sun4-sunos
839 Because the value of C<$Config{archname}> may depend on the
840 hardware architecture, it can vary more than the value of C<$^O>.
842 =head2 DOS and Derivatives
844 Perl has long been ported to Intel-style microcomputers running under
845 systems like PC-DOS, MS-DOS, OS/2, and most Windows platforms you can
846 bring yourself to mention (except for Windows CE, if you count that).
847 Users familiar with I<COMMAND.COM> or I<CMD.EXE> style shells should
848 be aware that each of these file specifications may have subtle
851 my $filespec0 = "c:/foo/bar/file.txt";
852 my $filespec1 = "c:\\foo\\bar\\file.txt";
853 my $filespec2 = 'c:\foo\bar\file.txt';
854 my $filespec3 = 'c:\\foo\\bar\\file.txt';
856 System calls accept either C</> or C<\> as the path separator.
857 However, many command-line utilities of DOS vintage treat C</> as
858 the option prefix, so may get confused by filenames containing C</>.
859 Aside from calling any external programs, C</> will work just fine,
860 and probably better, as it is more consistent with popular usage,
861 and avoids the problem of remembering what to backwhack and what
864 The DOS FAT filesystem can accommodate only "8.3" style filenames. Under
865 the "case-insensitive, but case-preserving" HPFS (OS/2) and NTFS (NT)
866 filesystems you may have to be careful about case returned with functions
867 like C<readdir> or used with functions like C<open> or C<opendir>.
869 DOS also treats several filenames as special, such as AUX, PRN,
870 NUL, CON, COM1, LPT1, LPT2, etc. Unfortunately, sometimes these
871 filenames won't even work if you include an explicit directory
872 prefix. It is best to avoid such filenames, if you want your code
873 to be portable to DOS and its derivatives. It's hard to know what
874 these all are, unfortunately.
876 Users of these operating systems may also wish to make use of
877 scripts such as I<pl2bat.bat> or I<pl2cmd> to
878 put wrappers around your scripts.
880 Newline (C<\n>) is translated as C<\015\012> by STDIO when reading from
881 and writing to files (see L<"Newlines">). C<binmode(FILEHANDLE)>
882 will keep C<\n> translated as C<\012> for that filehandle. Since it is a
883 no-op on other systems, C<binmode> should be used for cross-platform code
884 that deals with binary data. That's assuming you realize in advance
885 that your data is in binary. General-purpose programs should
886 often assume nothing about their data.
888 The C<$^O> variable and the C<$Config{archname}> values for various
889 DOSish perls are as follows:
891 OS $^O $Config{archname} ID Version
892 --------------------------------------------------------
896 Windows 3.1 ? ? 0 3 01
897 Windows 95 MSWin32 MSWin32-x86 1 4 00
898 Windows 98 MSWin32 MSWin32-x86 1 4 10
899 Windows ME MSWin32 MSWin32-x86 1 ?
900 Windows NT MSWin32 MSWin32-x86 2 4 xx
901 Windows NT MSWin32 MSWin32-ALPHA 2 4 xx
902 Windows NT MSWin32 MSWin32-ppc 2 4 xx
903 Windows 2000 MSWin32 MSWin32-x86 2 5 00
904 Windows XP MSWin32 MSWin32-x86 2 5 01
905 Windows 2003 MSWin32 MSWin32-x86 2 5 02
906 Windows Vista MSWin32 MSWin32-x86 2 6 00
907 Windows 7 MSWin32 MSWin32-x86 2 6 01
908 Windows 7 MSWin32 MSWin32-x64 2 6 01
909 Windows 2008 MSWin32 MSWin32-x86 2 6 01
910 Windows 2008 MSWin32 MSWin32-x64 2 6 01
911 Windows CE MSWin32 ? 3
914 The various MSWin32 Perl's can distinguish the OS they are running on
915 via the value of the fifth element of the list returned from
916 Win32::GetOSVersion(). For example:
918 if ($^O eq 'MSWin32') {
919 my @os_version_info = Win32::GetOSVersion();
920 print +('3.1','95','NT')[$os_version_info[4]],"\n";
923 There are also Win32::IsWinNT() and Win32::IsWin95(), try C<perldoc Win32>,
924 and as of libwin32 0.19 (not part of the core Perl distribution)
925 Win32::GetOSName(). The very portable POSIX::uname() will work too:
927 c:\> perl -MPOSIX -we "print join '|', uname"
928 Windows NT|moonru|5.0|Build 2195 (Service Pack 2)|x86
936 The djgpp environment for DOS, L<http://www.delorie.com/djgpp/>
941 The EMX environment for DOS, OS/2, etc. emx@iaehv.nl,
942 L<ftp://hobbes.nmsu.edu/pub/os2/dev/emx/> Also L<perlos2>.
946 Build instructions for Win32 in L<perlwin32>, or under the Cygnus environment
951 The C<Win32::*> modules in L<Win32>.
955 The ActiveState Pages, L<http://www.activestate.com/>
959 The Cygwin environment for Win32; F<README.cygwin> (installed
960 as L<perlcygwin>), L<http://www.cygwin.com/>
964 The U/WIN environment for Win32,
965 L<http://www.research.att.com/sw/tools/uwin/>
969 Build instructions for OS/2, L<perlos2>
975 Perl on VMS is discussed in L<perlvms> in the perl distribution.
977 The official name of VMS as of this writing is OpenVMS.
979 Perl on VMS can accept either VMS- or Unix-style file
980 specifications as in either of the following:
982 $ perl -ne "print if /perl_setup/i" SYS$LOGIN:LOGIN.COM
983 $ perl -ne "print if /perl_setup/i" /sys$login/login.com
985 but not a mixture of both as in:
987 $ perl -ne "print if /perl_setup/i" sys$login:/login.com
988 Can't open sys$login:/login.com: file specification syntax error
990 Interacting with Perl from the Digital Command Language (DCL) shell
991 often requires a different set of quotation marks than Unix shells do.
994 $ perl -e "print ""Hello, world.\n"""
997 There are several ways to wrap your perl scripts in DCL F<.COM> files, if
998 you are so inclined. For example:
1000 $ write sys$output "Hello from DCL!"
1002 $ then perl -x 'f$environment("PROCEDURE")
1003 $ else perl -x - 'p1 'p2 'p3 'p4 'p5 'p6 'p7 'p8
1004 $ deck/dollars="__END__"
1007 print "Hello from Perl!\n";
1012 Do take care with C<$ ASSIGN/nolog/user SYS$COMMAND: SYS$INPUT> if your
1013 perl-in-DCL script expects to do things like C<< $read = <STDIN>; >>.
1015 The VMS operating system has two filesystems, known as ODS-2 and ODS-5.
1017 For ODS-2, filenames are in the format "name.extension;version". The
1018 maximum length for filenames is 39 characters, and the maximum length for
1019 extensions is also 39 characters. Version is a number from 1 to
1020 32767. Valid characters are C</[A-Z0-9$_-]/>.
1022 The ODS-2 filesystem is case-insensitive and does not preserve case.
1023 Perl simulates this by converting all filenames to lowercase internally.
1025 For ODS-5, filenames may have almost any character in them and can include
1026 Unicode characters. Characters that could be misinterpreted by the DCL
1027 shell or file parsing utilities need to be prefixed with the C<^>
1028 character, or replaced with hexadecimal characters prefixed with the
1029 C<^> character. Such prefixing is only needed with the pathnames are
1030 in VMS format in applications. Programs that can accept the Unix format
1031 of pathnames do not need the escape characters. The maximum length for
1032 filenames is 255 characters. The ODS-5 file system can handle both
1033 a case preserved and a case sensitive mode.
1035 ODS-5 is only available on the OpenVMS for 64 bit platforms.
1037 Support for the extended file specifications is being done as optional
1038 settings to preserve backward compatibility with Perl scripts that
1039 assume the previous VMS limitations.
1041 In general routines on VMS that get a Unix format file specification
1042 should return it in a Unix format, and when they get a VMS format
1043 specification they should return a VMS format unless they are documented
1046 For routines that generate return a file specification, VMS allows setting
1047 if the C library which Perl is built on if it will be returned in VMS
1048 format or in Unix format.
1050 With the ODS-2 file system, there is not much difference in syntax of
1051 filenames without paths for VMS or Unix. With the extended character
1052 set available with ODS-5 there can be a significant difference.
1054 Because of this, existing Perl scripts written for VMS were sometimes
1055 treating VMS and Unix filenames interchangeably. Without the extended
1056 character set enabled, this behavior will mostly be maintained for
1057 backwards compatibility.
1059 When extended characters are enabled with ODS-5, the handling of
1060 Unix formatted file specifications is to that of a Unix system.
1062 VMS file specifications without extensions have a trailing dot. An
1063 equivalent Unix file specification should not show the trailing dot.
1065 The result of all of this, is that for VMS, for portable scripts, you
1066 can not depend on Perl to present the filenames in lowercase, to be
1067 case sensitive, and that the filenames could be returned in either
1070 And if a routine returns a file specification, unless it is intended to
1071 convert it, it should return it in the same format as it found it.
1073 C<readdir> by default has traditionally returned lowercased filenames.
1074 When the ODS-5 support is enabled, it will return the exact case of the
1075 filename on the disk.
1077 Files without extensions have a trailing period on them, so doing a
1078 C<readdir> in the default mode with a file named F<A.;5> will
1079 return F<a.> when VMS is (though that file could be opened with
1082 With support for extended file specifications and if C<opendir> was
1083 given a Unix format directory, a file named F<A.;5> will return F<a>
1084 and optionally in the exact case on the disk. When C<opendir> is given
1085 a VMS format directory, then C<readdir> should return F<a.>, and
1086 again with the optionally the exact case.
1088 RMS had an eight level limit on directory depths from any rooted logical
1089 (allowing 16 levels overall) prior to VMS 7.2, and even with versions of
1090 VMS on VAX up through 7.3. Hence C<PERL_ROOT:[LIB.2.3.4.5.6.7.8]> is a
1091 valid directory specification but C<PERL_ROOT:[LIB.2.3.4.5.6.7.8.9]> is
1092 not. F<Makefile.PL> authors might have to take this into account, but at
1093 least they can refer to the former as C</PERL_ROOT/lib/2/3/4/5/6/7/8/>.
1095 Pumpkings and module integrators can easily see whether files with too many
1096 directory levels have snuck into the core by running the following in the
1097 top-level source directory:
1099 $ perl -ne "$_=~s/\s+.*//; print if scalar(split /\//) > 8;" < MANIFEST
1102 The VMS::Filespec module, which gets installed as part of the build
1103 process on VMS, is a pure Perl module that can easily be installed on
1104 non-VMS platforms and can be helpful for conversions to and from RMS
1105 native formats. It is also now the only way that you should check to
1106 see if VMS is in a case sensitive mode.
1108 What C<\n> represents depends on the type of file opened. It usually
1109 represents C<\012> but it could also be C<\015>, C<\012>, C<\015\012>,
1110 C<\000>, C<\040>, or nothing depending on the file organization and
1111 record format. The VMS::Stdio module provides access to the
1112 special fopen() requirements of files with unusual attributes on VMS.
1114 TCP/IP stacks are optional on VMS, so socket routines might not be
1115 implemented. UDP sockets may not be supported.
1117 The TCP/IP library support for all current versions of VMS is dynamically
1118 loaded if present, so even if the routines are configured, they may
1119 return a status indicating that they are not implemented.
1121 The value of C<$^O> on OpenVMS is "VMS". To determine the architecture
1122 that you are running on without resorting to loading all of C<%Config>
1123 you can examine the content of the C<@INC> array like so:
1125 if (grep(/VMS_AXP/, @INC)) {
1126 print "I'm on Alpha!\n";
1128 } elsif (grep(/VMS_VAX/, @INC)) {
1129 print "I'm on VAX!\n";
1131 } elsif (grep(/VMS_IA64/, @INC)) {
1132 print "I'm on IA64!\n";
1135 print "I'm not so sure about where $^O is...\n";
1138 In general, the significant differences should only be if Perl is running
1139 on VMS_VAX or one of the 64 bit OpenVMS platforms.
1141 On VMS, perl determines the UTC offset from the C<SYS$TIMEZONE_DIFFERENTIAL>
1142 logical name. Although the VMS epoch began at 17-NOV-1858 00:00:00.00,
1143 calls to C<localtime> are adjusted to count offsets from
1144 01-JAN-1970 00:00:00.00, just like Unix.
1152 F<README.vms> (installed as F<README_vms>), L<perlvms>
1156 vmsperl list, vmsperl-subscribe@perl.org
1160 vmsperl on the web, L<http://www.sidhe.org/vmsperl/index.html>
1166 Perl on VOS (also known as OpenVOS) is discussed in F<README.vos>
1167 in the perl distribution (installed as L<perlvos>). Perl on VOS
1168 can accept either VOS- or Unix-style file specifications as in
1169 either of the following:
1171 $ perl -ne "print if /perl_setup/i" >system>notices
1172 $ perl -ne "print if /perl_setup/i" /system/notices
1174 or even a mixture of both as in:
1176 $ perl -ne "print if /perl_setup/i" >system/notices
1178 Even though VOS allows the slash character to appear in object
1179 names, because the VOS port of Perl interprets it as a pathname
1180 delimiting character, VOS files, directories, or links whose
1181 names contain a slash character cannot be processed. Such files
1182 must be renamed before they can be processed by Perl.
1184 Older releases of VOS (prior to OpenVOS Release 17.0) limit file
1185 names to 32 or fewer characters, prohibit file names from
1186 starting with a C<-> character, and prohibit file names from
1187 containing any character matching C<< tr/ !#%&'()*;<=>?// >>.
1189 Newer releases of VOS (OpenVOS Release 17.0 or later) support a
1190 feature known as extended names. On these releases, file names
1191 can contain up to 255 characters, are prohibited from starting
1192 with a C<-> character, and the set of prohibited characters is
1193 reduced to any character matching C<< tr/#%*<>?// >>. There are
1194 restrictions involving spaces and apostrophes: these characters
1195 must not begin or end a name, nor can they immediately precede or
1196 follow a period. Additionally, a space must not immediately
1197 precede another space or hyphen. Specifically, the following
1198 character combinations are prohibited: space-space,
1199 space-hyphen, period-space, space-period, period-apostrophe,
1200 apostrophe-period, leading or trailing space, and leading or
1201 trailing apostrophe. Although an extended file name is limited
1202 to 255 characters, a path name is still limited to 256
1205 The value of C<$^O> on VOS is "vos". To determine the
1206 architecture that you are running on without resorting to loading
1207 all of C<%Config> you can examine the content of the @INC array
1211 print "I'm on a Stratus box!\n";
1213 print "I'm not on a Stratus box!\n";
1223 F<README.vos> (installed as L<perlvos>)
1227 The VOS mailing list.
1229 There is no specific mailing list for Perl on VOS. You can contact
1230 the Stratus Technologies Customer Assistance Center (CAC) for your
1231 region, or you can use the contact information located in the
1232 distribution files on the Stratus Anonymous FTP site.
1236 Stratus Technologies on the web at L<http://www.stratus.com>
1240 VOS Open-Source Software on the web at L<http://ftp.stratus.com/pub/vos/vos.html>
1244 =head2 EBCDIC Platforms
1246 Recent versions of Perl have been ported to platforms such as OS/400 on
1247 AS/400 minicomputers as well as OS/390, VM/ESA, and BS2000 for S/390
1248 Mainframes. Such computers use EBCDIC character sets internally (usually
1249 Character Code Set ID 0037 for OS/400 and either 1047 or POSIX-BC for S/390
1250 systems). On the mainframe perl currently works under the "Unix system
1251 services for OS/390" (formerly known as OpenEdition), VM/ESA OpenEdition, or
1252 the BS200 POSIX-BC system (BS2000 is supported in perl 5.6 and greater).
1253 See L<perlos390> for details. Note that for OS/400 there is also a port of
1254 Perl 5.8.1/5.10.0 or later to the PASE which is ASCII-based (as opposed to
1255 ILE which is EBCDIC-based), see L<perlos400>.
1257 As of R2.5 of USS for OS/390 and Version 2.3 of VM/ESA these Unix
1258 sub-systems do not support the C<#!> shebang trick for script invocation.
1259 Hence, on OS/390 and VM/ESA perl scripts can be executed with a header
1260 similar to the following simple script:
1263 eval 'exec /usr/local/bin/perl -S $0 ${1+"$@"}'
1265 #!/usr/local/bin/perl # just a comment really
1267 print "Hello from perl!\n";
1269 OS/390 will support the C<#!> shebang trick in release 2.8 and beyond.
1270 Calls to C<system> and backticks can use POSIX shell syntax on all
1273 On the AS/400, if PERL5 is in your library list, you may need
1274 to wrap your perl scripts in a CL procedure to invoke them like so:
1277 CALL PGM(PERL5/PERL) PARM('/QOpenSys/hello.pl')
1280 This will invoke the perl script F<hello.pl> in the root of the
1281 QOpenSys file system. On the AS/400 calls to C<system> or backticks
1284 On these platforms, bear in mind that the EBCDIC character set may have
1285 an effect on what happens with some perl functions (such as C<chr>,
1286 C<pack>, C<print>, C<printf>, C<ord>, C<sort>, C<sprintf>, C<unpack>), as
1287 well as bit-fiddling with ASCII constants using operators like C<^>, C<&>
1288 and C<|>, not to mention dealing with socket interfaces to ASCII computers
1289 (see L<"Newlines">).
1291 Fortunately, most web servers for the mainframe will correctly
1292 translate the C<\n> in the following statement to its ASCII equivalent
1293 (C<\r> is the same under both Unix and OS/390):
1295 print "Content-type: text/html\r\n\r\n";
1297 The values of C<$^O> on some of these platforms includes:
1299 uname $^O $Config{'archname'}
1300 --------------------------------------------
1303 POSIX-BC posix-bc BS2000-posix-bc
1305 Some simple tricks for determining if you are running on an EBCDIC
1306 platform could include any of the following (perhaps all):
1308 if ("\t" eq "\005") { print "EBCDIC may be spoken here!\n"; }
1310 if (ord('A') == 193) { print "EBCDIC may be spoken here!\n"; }
1312 if (chr(169) eq 'z') { print "EBCDIC may be spoken here!\n"; }
1314 One thing you may not want to rely on is the EBCDIC encoding
1315 of punctuation characters since these may differ from code page to code
1316 page (and once your module or script is rumoured to work with EBCDIC,
1317 folks will want it to work with all EBCDIC character sets).
1325 L<perlos390>, F<README.os390>, F<perlbs2000>, L<perlebcdic>.
1329 The perl-mvs@perl.org list is for discussion of porting issues as well as
1330 general usage issues for all EBCDIC Perls. Send a message body of
1331 "subscribe perl-mvs" to majordomo@perl.org.
1335 AS/400 Perl information at
1336 L<http://as400.rochester.ibm.com/>
1337 as well as on CPAN in the F<ports/> directory.
1341 =head2 Acorn RISC OS
1343 Because Acorns use ASCII with newlines (C<\n>) in text files as C<\012> like
1344 Unix, and because Unix filename emulation is turned on by default,
1345 most simple scripts will probably work "out of the box". The native
1346 filesystem is modular, and individual filesystems are free to be
1347 case-sensitive or insensitive, and are usually case-preserving. Some
1348 native filesystems have name length limits, which file and directory
1349 names are silently truncated to fit. Scripts should be aware that the
1350 standard filesystem currently has a name length limit of B<10>
1351 characters, with up to 77 items in a directory, but other filesystems
1352 may not impose such limitations.
1354 Native filenames are of the form
1356 Filesystem#Special_Field::DiskName.$.Directory.Directory.File
1360 Special_Field is not usually present, but may contain . and $ .
1361 Filesystem =~ m|[A-Za-z0-9_]|
1362 DsicName =~ m|[A-Za-z0-9_/]|
1363 $ represents the root directory
1364 . is the path separator
1365 @ is the current directory (per filesystem but machine global)
1366 ^ is the parent directory
1367 Directory and File =~ m|[^\0- "\.\$\%\&:\@\\^\|\177]+|
1369 The default filename translation is roughly C<tr|/.|./|;>
1371 Note that C<"ADFS::HardDisk.$.File" ne 'ADFS::HardDisk.$.File'> and that
1372 the second stage of C<$> interpolation in regular expressions will fall
1373 foul of the C<$.> if scripts are not careful.
1375 Logical paths specified by system variables containing comma-separated
1376 search lists are also allowed; hence C<System:Modules> is a valid
1377 filename, and the filesystem will prefix C<Modules> with each section of
1378 C<System$Path> until a name is made that points to an object on disk.
1379 Writing to a new file C<System:Modules> would be allowed only if
1380 C<System$Path> contains a single item list. The filesystem will also
1381 expand system variables in filenames if enclosed in angle brackets, so
1382 C<< <System$Dir>.Modules >> would look for the file
1383 S<C<$ENV{'System$Dir'} . 'Modules'>>. The obvious implication of this is
1384 that B<fully qualified filenames can start with C<< <> >>> and should
1385 be protected when C<open> is used for input.
1387 Because C<.> was in use as a directory separator and filenames could not
1388 be assumed to be unique after 10 characters, Acorn implemented the C
1389 compiler to strip the trailing C<.c> C<.h> C<.s> and C<.o> suffix from
1390 filenames specified in source code and store the respective files in
1391 subdirectories named after the suffix. Hence files are translated:
1394 C:foo.h C:h.foo (logical path variable)
1395 sys/os.h sys.h.os (C compiler groks Unix-speak)
1396 10charname.c c.10charname
1397 10charname.o o.10charname
1398 11charname_.c c.11charname (assuming filesystem truncates at 10)
1400 The Unix emulation library's translation of filenames to native assumes
1401 that this sort of translation is required, and it allows a user-defined list
1402 of known suffixes that it will transpose in this fashion. This may
1403 seem transparent, but consider that with these rules F<foo/bar/baz.h>
1404 and F<foo/bar/h/baz> both map to F<foo.bar.h.baz>, and that C<readdir> and
1405 C<glob> cannot and do not attempt to emulate the reverse mapping. Other
1406 C<.>'s in filenames are translated to C</>.
1408 As implied above, the environment accessed through C<%ENV> is global, and
1409 the convention is that program specific environment variables are of the
1410 form C<Program$Name>. Each filesystem maintains a current directory,
1411 and the current filesystem's current directory is the B<global> current
1412 directory. Consequently, sociable programs don't change the current
1413 directory but rely on full pathnames, and programs (and Makefiles) cannot
1414 assume that they can spawn a child process which can change the current
1415 directory without affecting its parent (and everyone else for that
1418 Because native operating system filehandles are global and are currently
1419 allocated down from 255, with 0 being a reserved value, the Unix emulation
1420 library emulates Unix filehandles. Consequently, you can't rely on
1421 passing C<STDIN>, C<STDOUT>, or C<STDERR> to your children.
1423 The desire of users to express filenames of the form
1424 C<< <Foo$Dir>.Bar >> on the command line unquoted causes problems,
1425 too: C<``> command output capture has to perform a guessing game. It
1426 assumes that a string C<< <[^<>]+\$[^<>]> >> is a
1427 reference to an environment variable, whereas anything else involving
1428 C<< < >> or C<< > >> is redirection, and generally manages to be 99%
1429 right. Of course, the problem remains that scripts cannot rely on any
1430 Unix tools being available, or that any tools found have Unix-like command
1433 Extensions and XS are, in theory, buildable by anyone using free
1434 tools. In practice, many don't, as users of the Acorn platform are
1435 used to binary distributions. MakeMaker does run, but no available
1436 make currently copes with MakeMaker's makefiles; even if and when
1437 this should be fixed, the lack of a Unix-like shell will cause
1438 problems with makefile rules, especially lines of the form C<cd
1439 sdbm && make all>, and anything using quoting.
1441 "S<RISC OS>" is the proper name for the operating system, but the value
1442 in C<$^O> is "riscos" (because we don't like shouting).
1446 Perl has been ported to many platforms that do not fit into any of
1447 the categories listed above. Some, such as AmigaOS,
1448 QNX, Plan 9, and VOS, have been well-integrated into the standard
1449 Perl source code kit. You may need to see the F<ports/> directory
1450 on CPAN for information, and possibly binaries, for the likes of:
1451 aos, Atari ST, lynxos, riscos, Novell Netware, Tandem Guardian,
1452 I<etc.> (Yes, we know that some of these OSes may fall under the
1453 Unix category, but we are not a standards body.)
1455 Some approximate operating system names and their C<$^O> values
1456 in the "OTHER" category include:
1458 OS $^O $Config{'archname'}
1459 ------------------------------------------
1460 Amiga DOS amigaos m68k-amigos
1468 Amiga, F<README.amiga> (installed as L<perlamiga>).
1472 A free perl5-based PERL.NLM for Novell Netware is available in
1473 precompiled binary and source code form from L<http://www.novell.com/>
1474 as well as from CPAN.
1478 S<Plan 9>, F<README.plan9>
1482 =head1 FUNCTION IMPLEMENTATIONS
1484 Listed below are functions that are either completely unimplemented
1485 or else have been implemented differently on various platforms.
1486 Following each description will be, in parentheses, a list of
1487 platforms that the description applies to.
1489 The list may well be incomplete, or even wrong in some places. When
1490 in doubt, consult the platform-specific README files in the Perl
1491 source distribution, and any other documentation resources accompanying
1494 Be aware, moreover, that even among Unix-ish systems there are variations.
1496 For many functions, you can also query C<%Config>, exported by
1497 default from the Config module. For example, to check whether the
1498 platform has the C<lstat> call, check C<$Config{d_lstat}>. See
1499 L<Config> for a full description of available variables.
1501 =head2 Alphabetical Listing of Perl Functions
1507 C<-w> only inspects the read-only file attribute (FILE_ATTRIBUTE_READONLY),
1508 which determines whether the directory can be deleted, not whether it can
1509 be written to. Directories always have read and write access unless denied
1510 by discretionary access control lists (DACLs). (S<Win32>)
1512 C<-r>, C<-w>, C<-x>, and C<-o> tell whether the file is accessible,
1513 which may not reflect UIC-based file protections. (VMS)
1515 C<-s> by name on an open file will return the space reserved on disk,
1516 rather than the current extent. C<-s> on an open filehandle returns the
1517 current size. (S<RISC OS>)
1519 C<-R>, C<-W>, C<-X>, C<-O> are indistinguishable from C<-r>, C<-w>,
1520 C<-x>, C<-o>. (Win32, VMS, S<RISC OS>)
1522 C<-g>, C<-k>, C<-l>, C<-u>, C<-A> are not particularly meaningful.
1523 (Win32, VMS, S<RISC OS>)
1525 C<-p> is not particularly meaningful. (VMS, S<RISC OS>)
1527 C<-d> is true if passed a device spec without an explicit directory.
1530 C<-x> (or C<-X>) determine if a file ends in one of the executable
1531 suffixes. C<-S> is meaningless. (Win32)
1533 C<-x> (or C<-X>) determine if a file has an executable file type.
1538 Emulated using timers that must be explicitly polled whenever Perl
1539 wants to dispatch "safe signals" and therefore cannot interrupt
1540 blocking system calls. (Win32)
1544 Due to issues with various CPUs, math libraries, compilers, and standards,
1545 results for C<atan2()> may vary depending on any combination of the above.
1546 Perl attempts to conform to the Open Group/IEEE standards for the results
1547 returned from C<atan2()>, but cannot force the issue if the system Perl is
1548 run on does not allow it. (Tru64, HP-UX 10.20)
1550 The current version of the standards for C<atan2()> is available at
1551 L<http://www.opengroup.org/onlinepubs/009695399/functions/atan2.html>.
1555 Meaningless. (S<RISC OS>)
1557 Reopens file and restores pointer; if function fails, underlying
1558 filehandle may be closed, or pointer may be in a different position.
1561 The value returned by C<tell> may be affected after the call, and
1562 the filehandle may be flushed. (Win32)
1566 Only good for changing "owner" read-write access, "group", and "other"
1567 bits are meaningless. (Win32)
1569 Only good for changing "owner" and "other" read-write access. (S<RISC OS>)
1571 Access permissions are mapped onto VOS access-control list changes. (VOS)
1573 The actual permissions set depend on the value of the C<CYGWIN>
1574 in the SYSTEM environment settings. (Cygwin)
1576 Setting the exec bit on some locations (generally /sdcard) will return true
1577 but not actually set the bit. (Android)
1581 Not implemented. (Win32, S<Plan 9>, S<RISC OS>)
1583 Does nothing, but won't fail. (Win32)
1585 A little funky, because VOS's notion of ownership is a little funky (VOS).
1589 Not implemented. (Win32, VMS, S<Plan 9>, S<RISC OS>, VOS)
1593 May not be available if library or source was not provided when building
1596 Not implemented. (Android)
1600 Not implemented. (VMS, S<Plan 9>, VOS)
1604 Not implemented. (VMS, S<Plan 9>, VOS)
1608 Not useful. (S<RISC OS>)
1610 Not supported. (Cygwin, Win32)
1612 Invokes VMS debugger. (VMS)
1616 C<exec LIST> without the use of indirect object syntax (C<exec PROGRAM LIST>)
1617 may fall back to trying the shell if the first spawn() fails. (Win32)
1619 Does not automatically flush output handles on some platforms.
1620 (SunOS, Solaris, HP-UX)
1622 Not supported. (Symbian OS)
1626 Emulates Unix exit() (which considers C<exit 1> to indicate an error) by
1627 mapping the C<1> to SS$_ABORT (C<44>). This behavior may be overridden
1628 with the pragma C<use vmsish 'exit'>. As with the CRTL's exit()
1629 function, C<exit 0> is also mapped to an exit status of SS$_NORMAL
1630 (C<1>); this mapping cannot be overridden. Any other argument to exit()
1631 is used directly as Perl's exit status. On VMS, unless the future
1632 POSIX_EXIT mode is enabled, the exit code should always be a valid
1633 VMS exit code and not a generic number. When the POSIX_EXIT mode is
1634 enabled, a generic number will be encoded in a method compatible with
1635 the C library _POSIX_EXIT macro so that it can be decoded by other
1636 programs, particularly ones written in C, like the GNV package. (VMS)
1638 C<exit()> resets file pointers, which is a problem when called
1639 from a child process (created by C<fork()>) in C<BEGIN>.
1640 A workaround is to use C<POSIX::_exit>. (Solaris)
1642 exit unless $Config{archname} =~ /\bsolaris\b/;
1643 require POSIX and POSIX::_exit(0);
1647 Not implemented. (Win32)
1649 Some functions available based on the version of VMS. (VMS)
1653 Not implemented (VMS, S<RISC OS>, VOS).
1657 Not implemented. (AmigaOS, S<RISC OS>, VMS)
1659 Emulated using multiple interpreters. See L<perlfork>. (Win32)
1661 Does not automatically flush output handles on some platforms.
1662 (SunOS, Solaris, HP-UX)
1666 Not implemented. (S<RISC OS>)
1670 Not implemented. (Win32, VMS, S<RISC OS>)
1674 Not implemented. (Win32, S<RISC OS>)
1678 Not implemented. (Win32, VMS, S<RISC OS>, VOS)
1682 Not implemented. (Win32)
1684 Not useful. (S<RISC OS>)
1688 Not implemented. (Win32, VMS, S<RISC OS>)
1692 Not implemented. (Android, Win32, S<Plan 9>)
1696 Not implemented. (Win32)
1698 Not useful. (S<RISC OS>)
1702 Not implemented. (Win32, VMS, S<RISC OS>)
1706 Not implemented. (Android, Win32, S<Plan 9>)
1708 =item getprotobynumber
1710 Not implemented. (Android)
1716 Not implemented. (Android, Win32)
1720 Not implemented. (Android, Win32, VMS)
1724 C<gethostbyname('localhost')> does not work everywhere: you may have
1725 to use C<gethostbyname('127.0.0.1')>. (S<Irix 5>)
1729 Not implemented. (Win32)
1733 Not implemented. (Android, Win32, S<Plan 9>)
1737 Not implemented. (Android, Win32, S<Plan 9>)
1741 Not implemented. (Win32, S<Plan 9>)
1745 Not implemented. (Android)
1749 Not implemented. (Android, Win32, S<Plan 9>, S<RISC OS>)
1753 Not implemented. (Win32, S<Plan 9>, S<RISC OS>)
1757 Not implemented. (Android, Win32, S<Plan 9>, S<RISC OS>)
1761 Not implemented. (S<Plan 9>, Win32, S<RISC OS>)
1765 Not implemented. (Win32)
1767 Either not implemented or a no-op. (Android)
1771 Not implemented. (Android, S<RISC OS>, VMS, Win32)
1775 Not implemented. (Android, Win32)
1779 Not implemented. (Android, Win32, S<Plan 9>)
1783 Not implemented. (Android, Win32, S<Plan 9>)
1787 Not implemented. (S<Plan 9>, Win32)
1789 =item getsockopt SOCKET,LEVEL,OPTNAME
1791 Not implemented. (S<Plan 9>)
1795 This operator is implemented via the File::Glob extension on most
1796 platforms. See L<File::Glob> for portability information.
1800 In theory, gmtime() is reliable from -2**63 to 2**63-1. However,
1801 because work arounds in the implementation use floating point numbers,
1802 it will become inaccurate as the time gets larger. This is a bug and
1803 will be fixed in the future.
1805 On VOS, time values are 32-bit quantities.
1807 =item ioctl FILEHANDLE,FUNCTION,SCALAR
1809 Not implemented. (VMS)
1811 Available only for socket handles, and it does what the ioctlsocket() call
1812 in the Winsock API does. (Win32)
1814 Available only for socket handles. (S<RISC OS>)
1818 Not implemented, hence not useful for taint checking. (S<RISC OS>)
1820 C<kill()> doesn't have the semantics of C<raise()>, i.e. it doesn't send
1821 a signal to the identified process like it does on Unix platforms.
1822 Instead C<kill($sig, $pid)> terminates the process identified by $pid,
1823 and makes it exit immediately with exit status $sig. As in Unix, if
1824 $sig is 0 and the specified process exists, it returns true without
1825 actually terminating it. (Win32)
1827 C<kill(-9, $pid)> will terminate the process specified by $pid and
1828 recursively all child processes owned by it. This is different from
1829 the Unix semantics, where the signal will be delivered to all
1830 processes in the same process group as the process specified by
1833 Is not supported for process identification number of 0 or negative
1838 Not implemented. (S<RISC OS>, VOS)
1840 Link count not updated because hard links are not quite that hard
1841 (They are sort of half-way between hard and soft links). (AmigaOS)
1843 Hard links are implemented on Win32 under NTFS only. They are
1844 natively supported on Windows 2000 and later. On Windows NT they
1845 are implemented using the Windows POSIX subsystem support and the
1846 Perl process will need Administrator or Backup Operator privileges
1847 to create hard links.
1849 Available on 64 bit OpenVMS 8.2 and later. (VMS)
1853 localtime() has the same range as L</gmtime>, but because time zone
1854 rules change its accuracy for historical and future times may degrade
1855 but usually by no more than an hour.
1859 Not implemented. (S<RISC OS>)
1861 Return values (especially for device and inode) may be bogus. (Win32)
1871 Not implemented. (Android, Win32, VMS, S<Plan 9>, S<RISC OS>, VOS)
1875 open to C<|-> and C<-|> are unsupported. (Win32, S<RISC OS>)
1877 List-form pipe opens may fall back to the shell if the first spawn()
1880 Opening a process does not automatically flush output handles on some
1881 platforms. (SunOS, Solaris, HP-UX)
1885 Not implemented. (Win32, VMS, S<RISC OS>)
1889 Can't move directories between directories on different logical volumes. (Win32)
1893 Will not cause readdir() to re-read the directory stream. The entries
1894 already read before the rewinddir() call will just be returned again
1895 from a cache buffer. (Win32)
1899 Only implemented on sockets. (Win32, VMS)
1901 Only reliable on sockets. (S<RISC OS>)
1903 Note that the C<select FILEHANDLE> form is generally portable.
1911 Not implemented. (Android, Win32, VMS, S<RISC OS>)
1915 Not implemented. (Android, VMS, Win32, S<RISC OS>)
1919 Not implemented. (Win32, VMS, S<RISC OS>, VOS)
1923 Not implemented. (Win32, VMS, S<RISC OS>, VOS)
1927 Not implemented. (Android, Win32, S<RISC OS>)
1931 Not implemented. (S<Plan 9>)
1941 Not implemented. (Android, Win32, VMS, S<RISC OS>)
1945 Emulated using synchronization functions such that it can be
1946 interrupted by alarm(), and limited to a maximum of 4294967 seconds,
1947 approximately 49 days. (Win32)
1951 A relatively recent addition to socket functions, may not
1952 be implemented even in Unix platforms.
1956 Not implemented. (S<RISC OS>)
1958 Available on 64 bit OpenVMS 8.2 and later. (VMS)
1962 Platforms that do not have rdev, blksize, or blocks will return these
1963 as '', so numeric comparison or manipulation of these fields may cause
1964 'not numeric' warnings.
1966 ctime not supported on UFS (S<Mac OS X>).
1968 ctime is creation time instead of inode change time (Win32).
1970 device and inode are not meaningful. (Win32)
1972 device and inode are not necessarily reliable. (VMS)
1974 mtime, atime and ctime all return the last modification time. Device and
1975 inode are not necessarily reliable. (S<RISC OS>)
1977 dev, rdev, blksize, and blocks are not available. inode is not
1978 meaningful and will differ between stat calls on the same file. (os2)
1980 some versions of cygwin when doing a stat("foo") and if not finding it
1981 may then attempt to stat("foo.exe") (Cygwin)
1983 On Win32 stat() needs to open the file to determine the link count
1984 and update attributes that may have been changed through hard links.
1985 Setting ${^WIN32_SLOPPY_STAT} to a true value speeds up stat() by
1986 not performing this operation. (Win32)
1990 Not implemented. (Win32, S<RISC OS>)
1992 Implemented on 64 bit VMS 8.3. VMS requires the symbolic link to be in Unix
1993 syntax if it is intended to resolve to a valid path.
1997 Not implemented. (Win32, VMS, S<RISC OS>, VOS)
2001 The traditional "0", "1", and "2" MODEs are implemented with different
2002 numeric values on some systems. The flags exported by C<Fcntl>
2003 (O_RDONLY, O_WRONLY, O_RDWR) should work everywhere though. (S<Mac
2008 As an optimization, may not call the command shell specified in
2009 C<$ENV{PERL5SHELL}>. C<system(1, @args)> spawns an external
2010 process and immediately returns its process designator, without
2011 waiting for it to terminate. Return value may be used subsequently
2012 in C<wait> or C<waitpid>. Failure to spawn() a subprocess is indicated
2013 by setting $? to "255 << 8". C<$?> is set in a way compatible with
2014 Unix (i.e. the exitstatus of the subprocess is obtained by "$? >> 8",
2015 as described in the documentation). (Win32)
2017 There is no shell to process metacharacters, and the native standard is
2018 to pass a command line terminated by "\n" "\r" or "\0" to the spawned
2019 program. Redirection such as C<< > foo >> is performed (if at all) by
2020 the run time library of the spawned program. C<system> I<list> will call
2021 the Unix emulation library's C<exec> emulation, which attempts to provide
2022 emulation of the stdin, stdout, stderr in force in the parent, providing
2023 the child program uses a compatible version of the emulation library.
2024 I<scalar> will call the native command line direct and no such emulation
2025 of a child Unix program will exists. Mileage B<will> vary. (S<RISC OS>)
2027 C<system LIST> without the use of indirect object syntax (C<system PROGRAM LIST>)
2028 may fall back to trying the shell if the first spawn() fails. (Win32)
2030 Does not automatically flush output handles on some platforms.
2031 (SunOS, Solaris, HP-UX)
2033 The return value is POSIX-like (shifted up by 8 bits), which only allows
2034 room for a made-up value derived from the severity bits of the native
2035 32-bit condition code (unless overridden by C<use vmsish 'status'>).
2036 If the native condition code is one that has a POSIX value encoded, the
2037 POSIX value will be decoded to extract the expected exit value.
2038 For more details see L<perlvms/$?>. (VMS)
2042 Not implemented. (Android)
2046 "cumulative" times will be bogus. On anything other than Windows NT
2047 or Windows 2000, "system" time will be bogus, and "user" time is
2048 actually the time returned by the clock() function in the C runtime
2051 Not useful. (S<RISC OS>)
2055 Not implemented. (Older versions of VMS)
2057 Truncation to same-or-shorter lengths only. (VOS)
2059 If a FILEHANDLE is supplied, it must be writable and opened in append
2060 mode (i.e., use C<<< open(FH, '>>filename') >>>
2061 or C<sysopen(FH,...,O_APPEND|O_RDWR)>. If a filename is supplied, it
2062 should not be held open elsewhere. (Win32)
2066 Returns undef where unavailable.
2068 C<umask> works but the correct permissions are set only when the file
2069 is finally closed. (AmigaOS)
2073 Only the modification time is updated. (VMS, S<RISC OS>)
2075 May not behave as expected. Behavior depends on the C runtime
2076 library's implementation of utime(), and the filesystem being
2077 used. The FAT filesystem typically does not support an "access
2078 time" field, and it may limit timestamps to a granularity of
2079 two seconds. (Win32)
2085 Can only be applied to process handles returned for processes spawned
2086 using C<system(1, ...)> or pseudo processes created with C<fork()>. (Win32)
2088 Not useful. (S<RISC OS>)
2093 =head1 Supported Platforms
2095 The following platforms are known to build Perl 5.12 (as of April 2010,
2096 its release date) from the standard source code distribution available
2097 at L<http://www.cpan.org/src>
2101 =item Linux (x86, ARM, IA64)
2115 =item Windows Server 2003
2119 =item Windows Server 2008
2127 Some tests are known to fail:
2133 F<ext/XS-APItes/t/call_checker.t> - see
2134 L<https://rt.perl.org/Ticket/Display.html?id=78502>
2138 F<dist/I18N-Collate/t/I18N-Collate.t>
2142 F<ext/Win32CORE/t/win32core.t> - may fail on recent cygwin installs.
2146 =item Solaris (x86, SPARC)
2152 =item Alpha (7.2 and later)
2154 =item I64 (8.2 and later)
2164 =item Debian GNU/kFreeBSD
2168 =item Irix (6.5. What else?)
2176 =item QNX Neutrino RTOS (6.5.0)
2180 =item Stratus OpenVOS (17.0 or later)
2186 =item time_t issues that may or may not be fixed
2190 =item Symbian (Series 60 v3, 3.2 and 5 - what else?)
2192 =item Stratus VOS / OpenVOS
2200 Perl now builds with FreeMiNT/Atari. It fails a few tests, that needs
2203 The FreeMiNT port uses GNU dld for loadable module capabilities. So
2204 ensure you have that library installed when building perl.
2208 =head1 EOL Platforms
2212 The following platforms were supported by a previous version of
2213 Perl but have been officially removed from Perl's source code
2224 The following platforms were supported up to 5.10. They may still
2225 have worked in 5.12, but supporting code has been removed for 5.14:
2241 The following platforms were supported by a previous version of
2242 Perl but have been officially removed from Perl's source code
2249 =item Apollo Domain/OS
2251 =item Apple Mac OS 8/9
2258 =head1 Supported Platforms (Perl 5.8)
2260 As of July 2002 (the Perl release 5.8.0), the following platforms were
2261 able to build Perl from the standard source code distribution
2262 available at L<http://www.cpan.org/src/>
2273 HI-UXMPP (Hitachi) (5.8.0 worked but we didn't know it)
2283 ReliantUNIX (formerly SINIX)
2285 OpenVMS (formerly VMS)
2286 Open UNIX (Unixware) (since Perl 5.8.1/5.9.0)
2288 OS/400 (using the PASE) (since Perl 5.8.1/5.9.0)
2290 POSIX-BC (formerly BS2000)
2295 Tru64 UNIX (formerly DEC OSF/1, Digital UNIX)
2300 Win95/98/ME/2K/XP 2)
2302 z/OS (formerly OS/390)
2305 1) in DOS mode either the DOS or OS/2 ports can be used
2306 2) compilers: Borland, MinGW (GCC), VC6
2308 The following platforms worked with the previous releases (5.6 and
2309 5.7), but we did not manage either to fix or to test these in time
2310 for the 5.8.0 release. There is a very good chance that many of these
2311 will work fine with the 5.8.0.
2324 Known to be broken for 5.8.0 (but 5.6.1 and 5.7.2 can be used):
2328 The following platforms have been known to build Perl from source in
2329 the past (5.005_03 and earlier), but we haven't been able to verify
2330 their status for the current release, either because the
2331 hardware/software platforms are rare or because we don't have an
2332 active champion on these platforms--or both. They used to work,
2333 though, so go ahead and try compiling them, and let perlbug@perl.org
2366 The following platforms have their own source code distributions and
2367 binaries available via L<http://www.cpan.org/ports/>
2371 OS/400 (ILE) 5.005_02
2372 Tandem Guardian 5.004
2374 The following platforms have only binaries available via
2375 L<http://www.cpan.org/ports/index.html> :
2379 Acorn RISCOS 5.005_02
2383 Although we do suggest that you always build your own Perl from
2384 the source code, both for maximal configurability and for security,
2385 in case you are in a hurry you can check
2386 L<http://www.cpan.org/ports/index.html> for binary distributions.
2390 L<perlaix>, L<perlamiga>, L<perlbs2000>,
2391 L<perlce>, L<perlcygwin>, L<perldos>,
2392 L<perlebcdic>, L<perlfreebsd>, L<perlhurd>, L<perlhpux>, L<perlirix>,
2393 L<perlmacos>, L<perlmacosx>,
2394 L<perlnetware>, L<perlos2>, L<perlos390>, L<perlos400>,
2395 L<perlplan9>, L<perlqnx>, L<perlsolaris>, L<perltru64>,
2396 L<perlunicode>, L<perlvms>, L<perlvos>, L<perlwin32>, and L<Win32>.
2398 =head1 AUTHORS / CONTRIBUTORS
2400 Abigail <abigail@foad.org>,
2401 Charles Bailey <bailey@newman.upenn.edu>,
2402 Graham Barr <gbarr@pobox.com>,
2403 Tom Christiansen <tchrist@perl.com>,
2404 Nicholas Clark <nick@ccl4.org>,
2405 Thomas Dorner <Thomas.Dorner@start.de>,
2406 Andy Dougherty <doughera@lafayette.edu>,
2407 Dominic Dunlop <domo@computer.org>,
2408 Neale Ferguson <neale@vma.tabnsw.com.au>,
2409 David J. Fiander <davidf@mks.com>,
2410 Paul Green <Paul.Green@stratus.com>,
2411 M.J.T. Guy <mjtg@cam.ac.uk>,
2412 Jarkko Hietaniemi <jhi@iki.fi>,
2413 Luther Huffman <lutherh@stratcom.com>,
2414 Nick Ing-Simmons <nick@ing-simmons.net>,
2415 Andreas J. KE<ouml>nig <a.koenig@mind.de>,
2416 Markus Laker <mlaker@contax.co.uk>,
2417 Andrew M. Langmead <aml@world.std.com>,
2418 Larry Moore <ljmoore@freespace.net>,
2419 Paul Moore <Paul.Moore@uk.origin-it.com>,
2420 Chris Nandor <pudge@pobox.com>,
2421 Matthias Neeracher <neeracher@mac.com>,
2422 Philip Newton <pne@cpan.org>,
2423 Gary Ng <71564.1743@CompuServe.COM>,
2424 Tom Phoenix <rootbeer@teleport.com>,
2425 AndrE<eacute> Pirard <A.Pirard@ulg.ac.be>,
2426 Peter Prymmer <pvhp@forte.com>,
2427 Hugo van der Sanden <hv@crypt0.demon.co.uk>,
2428 Gurusamy Sarathy <gsar@activestate.com>,
2429 Paul J. Schinder <schinder@pobox.com>,
2430 Michael G Schwern <schwern@pobox.com>,
2431 Dan Sugalski <dan@sidhe.org>,
2432 Nathan Torkington <gnat@frii.com>,
2433 John Malmberg <wb8tyw@qsl.net>