1=head1 NAME 2 3perlport - Writing portable Perl 4 5=head1 DESCRIPTION 6 7Perl runs on numerous operating systems. While most of them share 8much in common, they also have their own unique features. 9 10This document is meant to help you to find out what constitutes portable 11Perl code. That way once you make a decision to write portably, 12you know where the lines are drawn, and you can stay within them. 13 14There is a tradeoff between taking full advantage of one particular 15type of computer and taking advantage of a full range of them. 16Naturally, as you broaden your range and become more diverse, the 17common factors drop, and you are left with an increasingly smaller 18area of common ground in which you can operate to accomplish a 19particular task. Thus, when you begin attacking a problem, it is 20important to consider under which part of the tradeoff curve you 21want to operate. Specifically, you must decide whether it is 22important that the task that you are coding have the full generality 23of being portable, or whether to just get the job done right now. 24This is the hardest choice to be made. The rest is easy, because 25Perl provides many choices, whichever way you want to approach your 26problem. 27 28Looking at it another way, writing portable code is usually about 29willfully limiting your available choices. Naturally, it takes 30discipline and sacrifice to do that. The product of portability 31and convenience may be a constant. You have been warned. 32 33Be aware of two important points: 34 35=over 4 36 37=item Not all Perl programs have to be portable 38 39There is no reason you should not use Perl as a language to glue Unix 40tools together, or to prototype a Macintosh application, or to manage the 41Windows registry. If it makes no sense to aim for portability for one 42reason or another in a given program, then don't bother. 43 44=item Nearly all of Perl already I<is> portable 45 46Don't be fooled into thinking that it is hard to create portable Perl 47code. It isn't. Perl tries its level-best to bridge the gaps between 48what's available on different platforms, and all the means available to 49use those features. Thus almost all Perl code runs on any machine 50without modification. But there are some significant issues in 51writing portable code, and this document is entirely about those issues. 52 53=back 54 55Here's the general rule: When you approach a task commonly done 56using a whole range of platforms, think about writing portable 57code. That way, you don't sacrifice much by way of the implementation 58choices you can avail yourself of, and at the same time you can give 59your users lots of platform choices. On the other hand, when you have to 60take advantage of some unique feature of a particular platform, as is 61often the case with systems programming (whether for Unix, Windows, 62S<Mac OS>, VMS, etc.), consider writing platform-specific code. 63 64When the code will run on only two or three operating systems, you 65may need to consider only the differences of those particular systems. 66The important thing is to decide where the code will run and to be 67deliberate in your decision. 68 69The material below is separated into three main sections: main issues of 70portability (L<"ISSUES">, platform-specific issues (L<"PLATFORMS">, and 71built-in perl functions that behave differently on various ports 72(L<"FUNCTION IMPLEMENTATIONS">. 73 74This information should not be considered complete; it includes possibly 75transient information about idiosyncrasies of some of the ports, almost 76all of which are in a state of constant evolution. Thus, this material 77should be considered a perpetual work in progress 78(C<< <IMG SRC="yellow_sign.gif" ALT="Under Construction"> >>). 79 80=head1 ISSUES 81 82=head2 Newlines 83 84In most operating systems, lines in files are terminated by newlines. 85Just what is used as a newline may vary from OS to OS. Unix 86traditionally uses C<\012>, one type of DOSish I/O uses C<\015\012>, 87and S<Mac OS> uses C<\015>. 88 89Perl uses C<\n> to represent the "logical" newline, where what is 90logical may depend on the platform in use. In MacPerl, C<\n> always 91means C<\015>. In DOSish perls, C<\n> usually means C<\012>, but 92when accessing a file in "text" mode, STDIO translates it to (or 93from) C<\015\012>, depending on whether you're reading or writing. 94Unix does the same thing on ttys in canonical mode. C<\015\012> 95is commonly referred to as CRLF. 96 97A common cause of unportable programs is the misuse of chop() to trim 98newlines: 99 100 # XXX UNPORTABLE! 101 while(<FILE>) { 102 chop; 103 @array = split(/:/); 104 #... 105 } 106 107You can get away with this on Unix and Mac OS (they have a single 108character end-of-line), but the same program will break under DOSish 109perls because you're only chop()ing half the end-of-line. Instead, 110chomp() should be used to trim newlines. The Dunce::Files module can 111help audit your code for misuses of chop(). 112 113When dealing with binary files (or text files in binary mode) be sure 114to explicitly set $/ to the appropriate value for your file format 115before using chomp(). 116 117Because of the "text" mode translation, DOSish perls have limitations 118in using C<seek> and C<tell> on a file accessed in "text" mode. 119Stick to C<seek>-ing to locations you got from C<tell> (and no 120others), and you are usually free to use C<seek> and C<tell> even 121in "text" mode. Using C<seek> or C<tell> or other file operations 122may be non-portable. If you use C<binmode> on a file, however, you 123can usually C<seek> and C<tell> with arbitrary values in safety. 124 125A common misconception in socket programming is that C<\n> eq C<\012> 126everywhere. When using protocols such as common Internet protocols, 127C<\012> and C<\015> are called for specifically, and the values of 128the logical C<\n> and C<\r> (carriage return) are not reliable. 129 130 print SOCKET "Hi there, client!\r\n"; # WRONG 131 print SOCKET "Hi there, client!\015\012"; # RIGHT 132 133However, using C<\015\012> (or C<\cM\cJ>, or C<\x0D\x0A>) can be tedious 134and unsightly, as well as confusing to those maintaining the code. As 135such, the Socket module supplies the Right Thing for those who want it. 136 137 use Socket qw(:DEFAULT :crlf); 138 print SOCKET "Hi there, client!$CRLF" # RIGHT 139 140When reading from a socket, remember that the default input record 141separator C<$/> is C<\n>, but robust socket code will recognize as 142either C<\012> or C<\015\012> as end of line: 143 144 while (<SOCKET>) { 145 # ... 146 } 147 148Because both CRLF and LF end in LF, the input record separator can 149be set to LF and any CR stripped later. Better to write: 150 151 use Socket qw(:DEFAULT :crlf); 152 local($/) = LF; # not needed if $/ is already \012 153 154 while (<SOCKET>) { 155 s/$CR?$LF/\n/; # not sure if socket uses LF or CRLF, OK 156 # s/\015?\012/\n/; # same thing 157 } 158 159This example is preferred over the previous one--even for Unix 160platforms--because now any C<\015>'s (C<\cM>'s) are stripped out 161(and there was much rejoicing). 162 163Similarly, functions that return text data--such as a function that 164fetches a web page--should sometimes translate newlines before 165returning the data, if they've not yet been translated to the local 166newline representation. A single line of code will often suffice: 167 168 $data =~ s/\015?\012/\n/g; 169 return $data; 170 171Some of this may be confusing. Here's a handy reference to the ASCII CR 172and LF characters. You can print it out and stick it in your wallet. 173 174 LF eq \012 eq \x0A eq \cJ eq chr(10) eq ASCII 10 175 CR eq \015 eq \x0D eq \cM eq chr(13) eq ASCII 13 176 177 | Unix | DOS | Mac | 178 --------------------------- 179 \n | LF | LF | CR | 180 \r | CR | CR | LF | 181 \n * | LF | CRLF | CR | 182 \r * | CR | CR | LF | 183 --------------------------- 184 * text-mode STDIO 185 186The Unix column assumes that you are not accessing a serial line 187(like a tty) in canonical mode. If you are, then CR on input becomes 188"\n", and "\n" on output becomes CRLF. 189 190These are just the most common definitions of C<\n> and C<\r> in Perl. 191There may well be others. For example, on an EBCDIC implementation 192such as z/OS (OS/390) or OS/400 (using the ILE, the PASE is ASCII-based) 193the above material is similar to "Unix" but the code numbers change: 194 195 LF eq \025 eq \x15 eq \cU eq chr(21) eq CP-1047 21 196 LF eq \045 eq \x25 eq chr(37) eq CP-0037 37 197 CR eq \015 eq \x0D eq \cM eq chr(13) eq CP-1047 13 198 CR eq \015 eq \x0D eq \cM eq chr(13) eq CP-0037 13 199 200 | z/OS | OS/400 | 201 ---------------------- 202 \n | LF | LF | 203 \r | CR | CR | 204 \n * | LF | LF | 205 \r * | CR | CR | 206 ---------------------- 207 * text-mode STDIO 208 209=head2 Numbers endianness and Width 210 211Different CPUs store integers and floating point numbers in different 212orders (called I<endianness>) and widths (32-bit and 64-bit being the 213most common today). This affects your programs when they attempt to transfer 214numbers in binary format from one CPU architecture to another, 215usually either "live" via network connection, or by storing the 216numbers to secondary storage such as a disk file or tape. 217 218Conflicting storage orders make utter mess out of the numbers. If a 219little-endian host (Intel, VAX) stores 0x12345678 (305419896 in 220decimal), a big-endian host (Motorola, Sparc, PA) reads it as 2210x78563412 (2018915346 in decimal). Alpha and MIPS can be either: 222Digital/Compaq used/uses them in little-endian mode; SGI/Cray uses 223them in big-endian mode. To avoid this problem in network (socket) 224connections use the C<pack> and C<unpack> formats C<n> and C<N>, the 225"network" orders. These are guaranteed to be portable. 226 227You can explore the endianness of your platform by unpacking a 228data structure packed in native format such as: 229 230 print unpack("h*", pack("s2", 1, 2)), "\n"; 231 # '10002000' on e.g. Intel x86 or Alpha 21064 in little-endian mode 232 # '00100020' on e.g. Motorola 68040 233 234If you need to distinguish between endian architectures you could use 235either of the variables set like so: 236 237 $is_big_endian = unpack("h*", pack("s", 1)) =~ /01/; 238 $is_little_endian = unpack("h*", pack("s", 1)) =~ /^1/; 239 240Differing widths can cause truncation even between platforms of equal 241endianness. The platform of shorter width loses the upper parts of the 242number. There is no good solution for this problem except to avoid 243transferring or storing raw binary numbers. 244 245One can circumnavigate both these problems in two ways. Either 246transfer and store numbers always in text format, instead of raw 247binary, or else consider using modules like Data::Dumper (included in 248the standard distribution as of Perl 5.005) and Storable (included as 249of perl 5.8). Keeping all data as text significantly simplifies matters. 250 251The v-strings are portable only up to v2147483647 (0x7FFFFFFF), that's 252how far EBCDIC, or more precisely UTF-EBCDIC will go. 253 254=head2 Files and Filesystems 255 256Most platforms these days structure files in a hierarchical fashion. 257So, it is reasonably safe to assume that all platforms support the 258notion of a "path" to uniquely identify a file on the system. How 259that path is really written, though, differs considerably. 260 261Although similar, file path specifications differ between Unix, 262Windows, S<Mac OS>, OS/2, VMS, VOS, S<RISC OS>, and probably others. 263Unix, for example, is one of the few OSes that has the elegant idea 264of a single root directory. 265 266DOS, OS/2, VMS, VOS, and Windows can work similarly to Unix with C</> 267as path separator, or in their own idiosyncratic ways (such as having 268several root directories and various "unrooted" device files such NIL: 269and LPT:). 270 271S<Mac OS> uses C<:> as a path separator instead of C</>. 272 273The filesystem may support neither hard links (C<link>) nor 274symbolic links (C<symlink>, C<readlink>, C<lstat>). 275 276The filesystem may support neither access timestamp nor change 277timestamp (meaning that about the only portable timestamp is the 278modification timestamp), or one second granularity of any timestamps 279(e.g. the FAT filesystem limits the time granularity to two seconds). 280 281The "inode change timestamp" (the C<-C> filetest) may really be the 282"creation timestamp" (which it is not in UNIX). 283 284VOS perl can emulate Unix filenames with C</> as path separator. The 285native pathname characters greater-than, less-than, number-sign, and 286percent-sign are always accepted. 287 288S<RISC OS> perl can emulate Unix filenames with C</> as path 289separator, or go native and use C<.> for path separator and C<:> to 290signal filesystems and disk names. 291 292Don't assume UNIX filesystem access semantics: that read, write, 293and execute are all the permissions there are, and even if they exist, 294that their semantics (for example what do r, w, and x mean on 295a directory) are the UNIX ones. The various UNIX/POSIX compatibility 296layers usually try to make interfaces like chmod() work, but sometimes 297there simply is no good mapping. 298 299If all this is intimidating, have no (well, maybe only a little) 300fear. There are modules that can help. The File::Spec modules 301provide methods to do the Right Thing on whatever platform happens 302to be running the program. 303 304 use File::Spec::Functions; 305 chdir(updir()); # go up one directory 306 $file = catfile(curdir(), 'temp', 'file.txt'); 307 # on Unix and Win32, './temp/file.txt' 308 # on Mac OS, ':temp:file.txt' 309 # on VMS, '[.temp]file.txt' 310 311File::Spec is available in the standard distribution as of version 3125.004_05. File::Spec::Functions is only in File::Spec 0.7 and later, 313and some versions of perl come with version 0.6. If File::Spec 314is not updated to 0.7 or later, you must use the object-oriented 315interface from File::Spec (or upgrade File::Spec). 316 317In general, production code should not have file paths hardcoded. 318Making them user-supplied or read from a configuration file is 319better, keeping in mind that file path syntax varies on different 320machines. 321 322This is especially noticeable in scripts like Makefiles and test suites, 323which often assume C</> as a path separator for subdirectories. 324 325Also of use is File::Basename from the standard distribution, which 326splits a pathname into pieces (base filename, full path to directory, 327and file suffix). 328 329Even when on a single platform (if you can call Unix a single platform), 330remember not to count on the existence or the contents of particular 331system-specific files or directories, like F</etc/passwd>, 332F</etc/sendmail.conf>, F</etc/resolv.conf>, or even F</tmp/>. For 333example, F</etc/passwd> may exist but not contain the encrypted 334passwords, because the system is using some form of enhanced security. 335Or it may not contain all the accounts, because the system is using NIS. 336If code does need to rely on such a file, include a description of the 337file and its format in the code's documentation, then make it easy for 338the user to override the default location of the file. 339 340Don't assume a text file will end with a newline. They should, 341but people forget. 342 343Do not have two files or directories of the same name with different 344case, like F<test.pl> and F<Test.pl>, as many platforms have 345case-insensitive (or at least case-forgiving) filenames. Also, try 346not to have non-word characters (except for C<.>) in the names, and 347keep them to the 8.3 convention, for maximum portability, onerous a 348burden though this may appear. 349 350Likewise, when using the AutoSplit module, try to keep your functions to 3518.3 naming and case-insensitive conventions; or, at the least, 352make it so the resulting files have a unique (case-insensitively) 353first 8 characters. 354 355Whitespace in filenames is tolerated on most systems, but not all, 356and even on systems where it might be tolerated, some utilities 357might become confused by such whitespace. 358 359Many systems (DOS, VMS) cannot have more than one C<.> in their filenames. 360 361Don't assume C<< > >> won't be the first character of a filename. 362Always use C<< < >> explicitly to open a file for reading, or even 363better, use the three-arg version of open, unless you want the user to 364be able to specify a pipe open. 365 366 open(FILE, '<', $existing_file) or die $!; 367 368If filenames might use strange characters, it is safest to open it 369with C<sysopen> instead of C<open>. C<open> is magic and can 370translate characters like C<< > >>, C<< < >>, and C<|>, which may 371be the wrong thing to do. (Sometimes, though, it's the right thing.) 372Three-arg open can also help protect against this translation in cases 373where it is undesirable. 374 375Don't use C<:> as a part of a filename since many systems use that for 376their own semantics (Mac OS Classic for separating pathname components, 377many networking schemes and utilities for separating the nodename and 378the pathname, and so on). For the same reasons, avoid C<@>, C<;> and 379C<|>. 380 381Don't assume that in pathnames you can collapse two leading slashes 382C<//> into one: some networking and clustering filesystems have special 383semantics for that. Let the operating system to sort it out. 384 385The I<portable filename characters> as defined by ANSI C are 386 387 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 388 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 389 0 1 2 3 4 5 6 7 8 9 390 . _ - 391 392and the "-" shouldn't be the first character. If you want to be 393hypercorrect, stay case-insensitive and within the 8.3 naming 394convention (all the files and directories have to be unique within one 395directory if their names are lowercased and truncated to eight 396characters before the C<.>, if any, and to three characters after the 397C<.>, if any). (And do not use C<.>s in directory names.) 398 399=head2 System Interaction 400 401Not all platforms provide a command line. These are usually platforms 402that rely primarily on a Graphical User Interface (GUI) for user 403interaction. A program requiring a command line interface might 404not work everywhere. This is probably for the user of the program 405to deal with, so don't stay up late worrying about it. 406 407Some platforms can't delete or rename files held open by the system, 408this limitation may also apply to changing filesystem metainformation 409like file permissions or owners. Remember to C<close> files when you 410are done with them. Don't C<unlink> or C<rename> an open file. Don't 411C<tie> or C<open> a file already tied or opened; C<untie> or C<close> 412it first. 413 414Don't open the same file more than once at a time for writing, as some 415operating systems put mandatory locks on such files. 416 417Don't assume that write/modify permission on a directory gives the 418right to add or delete files/directories in that directory. That is 419filesystem specific: in some filesystems you need write/modify 420permission also (or even just) in the file/directory itself. In some 421filesystems (AFS, DFS) the permission to add/delete directory entries 422is a completely separate permission. 423 424Don't assume that a single C<unlink> completely gets rid of the file: 425some filesystems (most notably the ones in VMS) have versioned 426filesystems, and unlink() removes only the most recent one (it doesn't 427remove all the versions because by default the native tools on those 428platforms remove just the most recent version, too). The portable 429idiom to remove all the versions of a file is 430 431 1 while unlink "file"; 432 433This will terminate if the file is undeleteable for some reason 434(protected, not there, and so on). 435 436Don't count on a specific environment variable existing in C<%ENV>. 437Don't count on C<%ENV> entries being case-sensitive, or even 438case-preserving. Don't try to clear %ENV by saying C<%ENV = ();>, or, 439if you really have to, make it conditional on C<$^O ne 'VMS'> since in 440VMS the C<%ENV> table is much more than a per-process key-value string 441table. 442 443Don't count on signals or C<%SIG> for anything. 444 445Don't count on filename globbing. Use C<opendir>, C<readdir>, and 446C<closedir> instead. 447 448Don't count on per-program environment variables, or per-program current 449directories. 450 451Don't count on specific values of C<$!>, neither numeric nor 452especially the strings values-- users may switch their locales causing 453error messages to be translated into their languages. If you can 454trust a POSIXish environment, you can portably use the symbols defined 455by the Errno module, like ENOENT. And don't trust on the values of C<$!> 456at all except immediately after a failed system call. 457 458=head2 Command names versus file pathnames 459 460Don't assume that the name used to invoke a command or program with 461C<system> or C<exec> can also be used to test for the existence of the 462file that holds the executable code for that command or program. 463First, many systems have "internal" commands that are built-in to the 464shell or OS and while these commands can be invoked, there is no 465corresponding file. Second, some operating systems (e.g., Cygwin, 466DJGPP, OS/2, and VOS) have required suffixes for executable files; 467these suffixes are generally permitted on the command name but are not 468required. Thus, a command like "perl" might exist in a file named 469"perl", "perl.exe", or "perl.pm", depending on the operating system. 470The variable "_exe" in the Config module holds the executable suffix, 471if any. Third, the VMS port carefully sets up $^X and 472$Config{perlpath} so that no further processing is required. This is 473just as well, because the matching regular expression used below would 474then have to deal with a possible trailing version number in the VMS 475file name. 476 477To convert $^X to a file pathname, taking account of the requirements 478of the various operating system possibilities, say: 479 use Config; 480 $thisperl = $^X; 481 if ($^O ne 'VMS') 482 {$thisperl .= $Config{_exe} unless $thisperl =~ m/$Config{_exe}$/i;} 483 484To convert $Config{perlpath} to a file pathname, say: 485 use Config; 486 $thisperl = $Config{perlpath}; 487 if ($^O ne 'VMS') 488 {$thisperl .= $Config{_exe} unless $thisperl =~ m/$Config{_exe}$/i;} 489 490=head2 Networking 491 492Don't assume that you can reach the public Internet. 493 494Don't assume that there is only one way to get through firewalls 495to the public Internet. 496 497Don't assume that you can reach outside world through any other port 498than 80, or some web proxy. ftp is blocked by many firewalls. 499 500Don't assume that you can send email by connecting to the local SMTP port. 501 502Don't assume that you can reach yourself or any node by the name 503'localhost'. The same goes for '127.0.0.1'. You will have to try both. 504 505Don't assume that the host has only one network card, or that it 506can't bind to many virtual IP addresses. 507 508Don't assume a particular network device name. 509 510Don't assume a particular set of ioctl()s will work. 511 512Don't assume that you can ping hosts and get replies. 513 514Don't assume that any particular port (service) will respond. 515 516Don't assume that Sys::Hostname() (or any other API or command) 517returns either a fully qualified hostname or a non-qualified hostname: 518it all depends on how the system had been configured. Also remember 519things like DHCP and NAT-- the hostname you get back might not be very 520useful. 521 522All the above "don't":s may look daunting, and they are -- but the key 523is to degrade gracefully if one cannot reach the particular network 524service one wants. Croaking or hanging do not look very professional. 525 526=head2 Interprocess Communication (IPC) 527 528In general, don't directly access the system in code meant to be 529portable. That means, no C<system>, C<exec>, C<fork>, C<pipe>, 530C<``>, C<qx//>, C<open> with a C<|>, nor any of the other things 531that makes being a perl hacker worth being. 532 533Commands that launch external processes are generally supported on 534most platforms (though many of them do not support any type of 535forking). The problem with using them arises from what you invoke 536them on. External tools are often named differently on different 537platforms, may not be available in the same location, might accept 538different arguments, can behave differently, and often present their 539results in a platform-dependent way. Thus, you should seldom depend 540on them to produce consistent results. (Then again, if you're calling 541I<netstat -a>, you probably don't expect it to run on both Unix and CP/M.) 542 543One especially common bit of Perl code is opening a pipe to B<sendmail>: 544 545 open(MAIL, '|/usr/lib/sendmail -t') 546 or die "cannot fork sendmail: $!"; 547 548This is fine for systems programming when sendmail is known to be 549available. But it is not fine for many non-Unix systems, and even 550some Unix systems that may not have sendmail installed. If a portable 551solution is needed, see the various distributions on CPAN that deal 552with it. Mail::Mailer and Mail::Send in the MailTools distribution are 553commonly used, and provide several mailing methods, including mail, 554sendmail, and direct SMTP (via Net::SMTP) if a mail transfer agent is 555not available. Mail::Sendmail is a standalone module that provides 556simple, platform-independent mailing. 557 558The Unix System V IPC (C<msg*(), sem*(), shm*()>) is not available 559even on all Unix platforms. 560 561Do not use either the bare result of C<pack("N", 10, 20, 30, 40)> or 562bare v-strings (such as C<v10.20.30.40>) to represent IPv4 addresses: 563both forms just pack the four bytes into network order. That this 564would be equal to the C language C<in_addr> struct (which is what the 565socket code internally uses) is not guaranteed. To be portable use 566the routines of the Socket extension, such as C<inet_aton()>, 567C<inet_ntoa()>, and C<sockaddr_in()>. 568 569The rule of thumb for portable code is: Do it all in portable Perl, or 570use a module (that may internally implement it with platform-specific 571code, but expose a common interface). 572 573=head2 External Subroutines (XS) 574 575XS code can usually be made to work with any platform, but dependent 576libraries, header files, etc., might not be readily available or 577portable, or the XS code itself might be platform-specific, just as Perl 578code might be. If the libraries and headers are portable, then it is 579normally reasonable to make sure the XS code is portable, too. 580 581A different type of portability issue arises when writing XS code: 582availability of a C compiler on the end-user's system. C brings 583with it its own portability issues, and writing XS code will expose 584you to some of those. Writing purely in Perl is an easier way to 585achieve portability. 586 587=head2 Standard Modules 588 589In general, the standard modules work across platforms. Notable 590exceptions are the CPAN module (which currently makes connections to external 591programs that may not be available), platform-specific modules (like 592ExtUtils::MM_VMS), and DBM modules. 593 594There is no one DBM module available on all platforms. 595SDBM_File and the others are generally available on all Unix and DOSish 596ports, but not in MacPerl, where only NBDM_File and DB_File are 597available. 598 599The good news is that at least some DBM module should be available, and 600AnyDBM_File will use whichever module it can find. Of course, then 601the code needs to be fairly strict, dropping to the greatest common 602factor (e.g., not exceeding 1K for each record), so that it will 603work with any DBM module. See L<AnyDBM_File> for more details. 604 605=head2 Time and Date 606 607The system's notion of time of day and calendar date is controlled in 608widely different ways. Don't assume the timezone is stored in C<$ENV{TZ}>, 609and even if it is, don't assume that you can control the timezone through 610that variable. Don't assume anything about the three-letter timezone 611abbreviations (for example that MST would be the Mountain Standard Time, 612it's been known to stand for Moscow Standard Time). If you need to 613use timezones, express them in some unambiguous format like the 614exact number of minutes offset from UTC, or the POSIX timezone 615format. 616 617Don't assume that the epoch starts at 00:00:00, January 1, 1970, 618because that is OS- and implementation-specific. It is better to 619store a date in an unambiguous representation. The ISO 8601 standard 620defines YYYY-MM-DD as the date format, or YYYY-MM-DDTHH-MM-SS 621(that's a literal "T" separating the date from the time). 622Please do use the ISO 8601 instead of making us to guess what 623date 02/03/04 might be. ISO 8601 even sorts nicely as-is. 624A text representation (like "1987-12-18") can be easily converted 625into an OS-specific value using a module like Date::Parse. 626An array of values, such as those returned by C<localtime>, can be 627converted to an OS-specific representation using Time::Local. 628 629When calculating specific times, such as for tests in time or date modules, 630it may be appropriate to calculate an offset for the epoch. 631 632 require Time::Local; 633 $offset = Time::Local::timegm(0, 0, 0, 1, 0, 70); 634 635The value for C<$offset> in Unix will be C<0>, but in Mac OS will be 636some large number. C<$offset> can then be added to a Unix time value 637to get what should be the proper value on any system. 638 639On Windows (at least), you shouldn't pass a negative value to C<gmtime> or 640C<localtime>. 641 642=head2 Character sets and character encoding 643 644Assume very little about character sets. 645 646Assume nothing about numerical values (C<ord>, C<chr>) of characters. 647Do not use explicit code point ranges (like \xHH-\xHH); use for 648example symbolic character classes like C<[:print:]>. 649 650Do not assume that the alphabetic characters are encoded contiguously 651(in the numeric sense). There may be gaps. 652 653Do not assume anything about the ordering of the characters. 654The lowercase letters may come before or after the uppercase letters; 655the lowercase and uppercase may be interlaced so that both `a' and `A' 656come before `b'; the accented and other international characters may 657be interlaced so that E<auml> comes before `b'. 658 659=head2 Internationalisation 660 661If you may assume POSIX (a rather large assumption), you may read 662more about the POSIX locale system from L<perllocale>. The locale 663system at least attempts to make things a little bit more portable, 664or at least more convenient and native-friendly for non-English 665users. The system affects character sets and encoding, and date 666and time formatting--amongst other things. 667 668If you really want to be international, you should consider Unicode. 669See L<perluniintro> and L<perlunicode> for more information. 670 671If you want to use non-ASCII bytes (outside the bytes 0x00..0x7f) in 672the "source code" of your code, to be portable you have to be explicit 673about what bytes they are. Someone might for example be using your 674code under a UTF-8 locale, in which case random native bytes might be 675illegal ("Malformed UTF-8 ...") This means that for example embedding 676ISO 8859-1 bytes beyond 0x7f into your strings might cause trouble 677later. If the bytes are native 8-bit bytes, you can use the C<bytes> 678pragma. If the bytes are in a string (regular expression being a 679curious string), you can often also use the C<\xHH> notation instead 680of embedding the bytes as-is. If they are in some particular legacy 681encoding (ether single-byte or something more complicated), you can 682use the C<encoding> pragma. (If you want to write your code in UTF-8, 683you can use either the C<utf8> pragma, or the C<encoding> pragma.) 684The C<bytes> and C<utf8> pragmata are available since Perl 5.6.0, and 685the C<encoding> pragma since Perl 5.8.0. 686 687=head2 System Resources 688 689If your code is destined for systems with severely constrained (or 690missing!) virtual memory systems then you want to be I<especially> mindful 691of avoiding wasteful constructs such as: 692 693 # NOTE: this is no longer "bad" in perl5.005 694 for (0..10000000) {} # bad 695 for (my $x = 0; $x <= 10000000; ++$x) {} # good 696 697 @lines = <VERY_LARGE_FILE>; # bad 698 699 while (<FILE>) {$file .= $_} # sometimes bad 700 $file = join('', <FILE>); # better 701 702The last two constructs may appear unintuitive to most people. The 703first repeatedly grows a string, whereas the second allocates a 704large chunk of memory in one go. On some systems, the second is 705more efficient that the first. 706 707=head2 Security 708 709Most multi-user platforms provide basic levels of security, usually 710implemented at the filesystem level. Some, however, do 711not-- unfortunately. Thus the notion of user id, or "home" directory, 712or even the state of being logged-in, may be unrecognizable on many 713platforms. If you write programs that are security-conscious, it 714is usually best to know what type of system you will be running 715under so that you can write code explicitly for that platform (or 716class of platforms). 717 718Don't assume the UNIX filesystem access semantics: the operating 719system or the filesystem may be using some ACL systems, which are 720richer languages than the usual rwx. Even if the rwx exist, 721their semantics might be different. 722 723(From security viewpoint testing for permissions before attempting to 724do something is silly anyway: if one tries this, there is potential 725for race conditions-- someone or something might change the 726permissions between the permissions check and the actual operation. 727Just try the operation.) 728 729Don't assume the UNIX user and group semantics: especially, don't 730expect the C<< $< >> and C<< $> >> (or the C<$(> and C<$)>) to work 731for switching identities (or memberships). 732 733Don't assume set-uid and set-gid semantics. (And even if you do, 734think twice: set-uid and set-gid are a known can of security worms.) 735 736=head2 Style 737 738For those times when it is necessary to have platform-specific code, 739consider keeping the platform-specific code in one place, making porting 740to other platforms easier. Use the Config module and the special 741variable C<$^O> to differentiate platforms, as described in 742L<"PLATFORMS">. 743 744Be careful in the tests you supply with your module or programs. 745Module code may be fully portable, but its tests might not be. This 746often happens when tests spawn off other processes or call external 747programs to aid in the testing, or when (as noted above) the tests 748assume certain things about the filesystem and paths. Be careful not 749to depend on a specific output style for errors, such as when checking 750C<$!> after a failed system call. Using C<$!> for anything else than 751displaying it as output is doubtful (though see the Errno module for 752testing reasonably portably for error value). Some platforms expect 753a certain output format, and Perl on those platforms may have been 754adjusted accordingly. Most specifically, don't anchor a regex when 755testing an error value. 756 757=head1 CPAN Testers 758 759Modules uploaded to CPAN are tested by a variety of volunteers on 760different platforms. These CPAN testers are notified by mail of each 761new upload, and reply to the list with PASS, FAIL, NA (not applicable to 762this platform), or UNKNOWN (unknown), along with any relevant notations. 763 764The purpose of the testing is twofold: one, to help developers fix any 765problems in their code that crop up because of lack of testing on other 766platforms; two, to provide users with information about whether 767a given module works on a given platform. 768 769=over 4 770 771=item Mailing list: cpan-testers@perl.org 772 773=item Testing results: http://testers.cpan.org/ 774 775=back 776 777=head1 PLATFORMS 778 779As of version 5.002, Perl is built with a C<$^O> variable that 780indicates the operating system it was built on. This was implemented 781to help speed up code that would otherwise have to C<use Config> 782and use the value of C<$Config{osname}>. Of course, to get more 783detailed information about the system, looking into C<%Config> is 784certainly recommended. 785 786C<%Config> cannot always be trusted, however, because it was built 787at compile time. If perl was built in one place, then transferred 788elsewhere, some values may be wrong. The values may even have been 789edited after the fact. 790 791=head2 Unix 792 793Perl works on a bewildering variety of Unix and Unix-like platforms (see 794e.g. most of the files in the F<hints/> directory in the source code kit). 795On most of these systems, the value of C<$^O> (hence C<$Config{'osname'}>, 796too) is determined either by lowercasing and stripping punctuation from the 797first field of the string returned by typing C<uname -a> (or a similar command) 798at the shell prompt or by testing the file system for the presence of 799uniquely named files such as a kernel or header file. Here, for example, 800are a few of the more popular Unix flavors: 801 802 uname $^O $Config{'archname'} 803 -------------------------------------------- 804 AIX aix aix 805 BSD/OS bsdos i386-bsdos 806 Darwin darwin darwin 807 dgux dgux AViiON-dgux 808 DYNIX/ptx dynixptx i386-dynixptx 809 FreeBSD freebsd freebsd-i386 810 Linux linux arm-linux 811 Linux linux i386-linux 812 Linux linux i586-linux 813 Linux linux ppc-linux 814 HP-UX hpux PA-RISC1.1 815 IRIX irix irix 816 Mac OS X darwin darwin 817 MachTen PPC machten powerpc-machten 818 NeXT 3 next next-fat 819 NeXT 4 next OPENSTEP-Mach 820 openbsd openbsd i386-openbsd 821 OSF1 dec_osf alpha-dec_osf 822 reliantunix-n svr4 RM400-svr4 823 SCO_SV sco_sv i386-sco_sv 824 SINIX-N svr4 RM400-svr4 825 sn4609 unicos CRAY_C90-unicos 826 sn6521 unicosmk t3e-unicosmk 827 sn9617 unicos CRAY_J90-unicos 828 SunOS solaris sun4-solaris 829 SunOS solaris i86pc-solaris 830 SunOS4 sunos sun4-sunos 831 832Because the value of C<$Config{archname}> may depend on the 833hardware architecture, it can vary more than the value of C<$^O>. 834 835=head2 DOS and Derivatives 836 837Perl has long been ported to Intel-style microcomputers running under 838systems like PC-DOS, MS-DOS, OS/2, and most Windows platforms you can 839bring yourself to mention (except for Windows CE, if you count that). 840Users familiar with I<COMMAND.COM> or I<CMD.EXE> style shells should 841be aware that each of these file specifications may have subtle 842differences: 843 844 $filespec0 = "c:/foo/bar/file.txt"; 845 $filespec1 = "c:\\foo\\bar\\file.txt"; 846 $filespec2 = 'c:\foo\bar\file.txt'; 847 $filespec3 = 'c:\\foo\\bar\\file.txt'; 848 849System calls accept either C</> or C<\> as the path separator. 850However, many command-line utilities of DOS vintage treat C</> as 851the option prefix, so may get confused by filenames containing C</>. 852Aside from calling any external programs, C</> will work just fine, 853and probably better, as it is more consistent with popular usage, 854and avoids the problem of remembering what to backwhack and what 855not to. 856 857The DOS FAT filesystem can accommodate only "8.3" style filenames. Under 858the "case-insensitive, but case-preserving" HPFS (OS/2) and NTFS (NT) 859filesystems you may have to be careful about case returned with functions 860like C<readdir> or used with functions like C<open> or C<opendir>. 861 862DOS also treats several filenames as special, such as AUX, PRN, 863NUL, CON, COM1, LPT1, LPT2, etc. Unfortunately, sometimes these 864filenames won't even work if you include an explicit directory 865prefix. It is best to avoid such filenames, if you want your code 866to be portable to DOS and its derivatives. It's hard to know what 867these all are, unfortunately. 868 869Users of these operating systems may also wish to make use of 870scripts such as I<pl2bat.bat> or I<pl2cmd> to 871put wrappers around your scripts. 872 873Newline (C<\n>) is translated as C<\015\012> by STDIO when reading from 874and writing to files (see L<"Newlines">). C<binmode(FILEHANDLE)> 875will keep C<\n> translated as C<\012> for that filehandle. Since it is a 876no-op on other systems, C<binmode> should be used for cross-platform code 877that deals with binary data. That's assuming you realize in advance 878that your data is in binary. General-purpose programs should 879often assume nothing about their data. 880 881The C<$^O> variable and the C<$Config{archname}> values for various 882DOSish perls are as follows: 883 884 OS $^O $Config{archname} ID Version 885 -------------------------------------------------------- 886 MS-DOS dos ? 887 PC-DOS dos ? 888 OS/2 os2 ? 889 Windows 3.1 ? ? 0 3 01 890 Windows 95 MSWin32 MSWin32-x86 1 4 00 891 Windows 98 MSWin32 MSWin32-x86 1 4 10 892 Windows ME MSWin32 MSWin32-x86 1 ? 893 Windows NT MSWin32 MSWin32-x86 2 4 xx 894 Windows NT MSWin32 MSWin32-ALPHA 2 4 xx 895 Windows NT MSWin32 MSWin32-ppc 2 4 xx 896 Windows 2000 MSWin32 MSWin32-x86 2 5 xx 897 Windows XP MSWin32 MSWin32-x86 2 ? 898 Windows CE MSWin32 ? 3 899 Cygwin cygwin ? 900 901The various MSWin32 Perl's can distinguish the OS they are running on 902via the value of the fifth element of the list returned from 903Win32::GetOSVersion(). For example: 904 905 if ($^O eq 'MSWin32') { 906 my @os_version_info = Win32::GetOSVersion(); 907 print +('3.1','95','NT')[$os_version_info[4]],"\n"; 908 } 909 910There are also Win32::IsWinNT() and Win32::IsWin95(), try C<perldoc Win32>, 911and as of libwin32 0.19 (not part of the core Perl distribution) 912Win32::GetOSName(). The very portable POSIX::uname() will work too: 913 914 c:\> perl -MPOSIX -we "print join '|', uname" 915 Windows NT|moonru|5.0|Build 2195 (Service Pack 2)|x86 916 917Also see: 918 919=over 4 920 921=item * 922 923The djgpp environment for DOS, http://www.delorie.com/djgpp/ 924and L<perldos>. 925 926=item * 927 928The EMX environment for DOS, OS/2, etc. emx@iaehv.nl, 929http://www.leo.org/pub/comp/os/os2/leo/gnu/emx+gcc/index.html or 930ftp://hobbes.nmsu.edu/pub/os2/dev/emx/ Also L<perlos2>. 931 932=item * 933 934Build instructions for Win32 in L<perlwin32>, or under the Cygnus environment 935in L<perlcygwin>. 936 937=item * 938 939The C<Win32::*> modules in L<Win32>. 940 941=item * 942 943The ActiveState Pages, http://www.activestate.com/ 944 945=item * 946 947The Cygwin environment for Win32; F<README.cygwin> (installed 948as L<perlcygwin>), http://www.cygwin.com/ 949 950=item * 951 952The U/WIN environment for Win32, 953http://www.research.att.com/sw/tools/uwin/ 954 955=item * 956 957Build instructions for OS/2, L<perlos2> 958 959=back 960 961=head2 S<Mac OS> 962 963Any module requiring XS compilation is right out for most people, because 964MacPerl is built using non-free (and non-cheap!) compilers. Some XS 965modules that can work with MacPerl are built and distributed in binary 966form on CPAN. 967 968Directories are specified as: 969 970 volume:folder:file for absolute pathnames 971 volume:folder: for absolute pathnames 972 :folder:file for relative pathnames 973 :folder: for relative pathnames 974 :file for relative pathnames 975 file for relative pathnames 976 977Files are stored in the directory in alphabetical order. Filenames are 978limited to 31 characters, and may include any character except for 979null and C<:>, which is reserved as the path separator. 980 981Instead of C<flock>, see C<FSpSetFLock> and C<FSpRstFLock> in the 982Mac::Files module, or C<chmod(0444, ...)> and C<chmod(0666, ...)>. 983 984In the MacPerl application, you can't run a program from the command line; 985programs that expect C<@ARGV> to be populated can be edited with something 986like the following, which brings up a dialog box asking for the command 987line arguments. 988 989 if (!@ARGV) { 990 @ARGV = split /\s+/, MacPerl::Ask('Arguments?'); 991 } 992 993A MacPerl script saved as a "droplet" will populate C<@ARGV> with the full 994pathnames of the files dropped onto the script. 995 996Mac users can run programs under a type of command line interface 997under MPW (Macintosh Programmer's Workshop, a free development 998environment from Apple). MacPerl was first introduced as an MPW 999tool, and MPW can be used like a shell: 1000 1001 perl myscript.plx some arguments 1002 1003ToolServer is another app from Apple that provides access to MPW tools 1004from MPW and the MacPerl app, which allows MacPerl programs to use 1005C<system>, backticks, and piped C<open>. 1006 1007"S<Mac OS>" is the proper name for the operating system, but the value 1008in C<$^O> is "MacOS". To determine architecture, version, or whether 1009the application or MPW tool version is running, check: 1010 1011 $is_app = $MacPerl::Version =~ /App/; 1012 $is_tool = $MacPerl::Version =~ /MPW/; 1013 ($version) = $MacPerl::Version =~ /^(\S+)/; 1014 $is_ppc = $MacPerl::Architecture eq 'MacPPC'; 1015 $is_68k = $MacPerl::Architecture eq 'Mac68K'; 1016 1017S<Mac OS X>, based on NeXT's OpenStep OS, runs MacPerl natively, under the 1018"Classic" environment. There is no "Carbon" version of MacPerl to run 1019under the primary Mac OS X environment. S<Mac OS X> and its Open Source 1020version, Darwin, both run Unix perl natively. 1021 1022Also see: 1023 1024=over 4 1025 1026=item * 1027 1028MacPerl Development, http://dev.macperl.org/ . 1029 1030=item * 1031 1032The MacPerl Pages, http://www.macperl.com/ . 1033 1034=item * 1035 1036The MacPerl mailing lists, http://lists.perl.org/ . 1037 1038=back 1039 1040=head2 VMS 1041 1042Perl on VMS is discussed in L<perlvms> in the perl distribution. 1043Perl on VMS can accept either VMS- or Unix-style file 1044specifications as in either of the following: 1045 1046 $ perl -ne "print if /perl_setup/i" SYS$LOGIN:LOGIN.COM 1047 $ perl -ne "print if /perl_setup/i" /sys$login/login.com 1048 1049but not a mixture of both as in: 1050 1051 $ perl -ne "print if /perl_setup/i" sys$login:/login.com 1052 Can't open sys$login:/login.com: file specification syntax error 1053 1054Interacting with Perl from the Digital Command Language (DCL) shell 1055often requires a different set of quotation marks than Unix shells do. 1056For example: 1057 1058 $ perl -e "print ""Hello, world.\n""" 1059 Hello, world. 1060 1061There are several ways to wrap your perl scripts in DCL F<.COM> files, if 1062you are so inclined. For example: 1063 1064 $ write sys$output "Hello from DCL!" 1065 $ if p1 .eqs. "" 1066 $ then perl -x 'f$environment("PROCEDURE") 1067 $ else perl -x - 'p1 'p2 'p3 'p4 'p5 'p6 'p7 'p8 1068 $ deck/dollars="__END__" 1069 #!/usr/bin/perl 1070 1071 print "Hello from Perl!\n"; 1072 1073 __END__ 1074 $ endif 1075 1076Do take care with C<$ ASSIGN/nolog/user SYS$COMMAND: SYS$INPUT> if your 1077perl-in-DCL script expects to do things like C<< $read = <STDIN>; >>. 1078 1079Filenames are in the format "name.extension;version". The maximum 1080length for filenames is 39 characters, and the maximum length for 1081extensions is also 39 characters. Version is a number from 1 to 108232767. Valid characters are C</[A-Z0-9$_-]/>. 1083 1084VMS's RMS filesystem is case-insensitive and does not preserve case. 1085C<readdir> returns lowercased filenames, but specifying a file for 1086opening remains case-insensitive. Files without extensions have a 1087trailing period on them, so doing a C<readdir> with a file named F<A.;5> 1088will return F<a.> (though that file could be opened with 1089C<open(FH, 'A')>). 1090 1091RMS had an eight level limit on directory depths from any rooted logical 1092(allowing 16 levels overall) prior to VMS 7.2. Hence 1093C<PERL_ROOT:[LIB.2.3.4.5.6.7.8]> is a valid directory specification but 1094C<PERL_ROOT:[LIB.2.3.4.5.6.7.8.9]> is not. F<Makefile.PL> authors might 1095have to take this into account, but at least they can refer to the former 1096as C</PERL_ROOT/lib/2/3/4/5/6/7/8/>. 1097 1098The VMS::Filespec module, which gets installed as part of the build 1099process on VMS, is a pure Perl module that can easily be installed on 1100non-VMS platforms and can be helpful for conversions to and from RMS 1101native formats. 1102 1103What C<\n> represents depends on the type of file opened. It usually 1104represents C<\012> but it could also be C<\015>, C<\012>, C<\015\012>, 1105C<\000>, C<\040>, or nothing depending on the file organiztion and 1106record format. The VMS::Stdio module provides access to the 1107special fopen() requirements of files with unusual attributes on VMS. 1108 1109TCP/IP stacks are optional on VMS, so socket routines might not be 1110implemented. UDP sockets may not be supported. 1111 1112The value of C<$^O> on OpenVMS is "VMS". To determine the architecture 1113that you are running on without resorting to loading all of C<%Config> 1114you can examine the content of the C<@INC> array like so: 1115 1116 if (grep(/VMS_AXP/, @INC)) { 1117 print "I'm on Alpha!\n"; 1118 1119 } elsif (grep(/VMS_VAX/, @INC)) { 1120 print "I'm on VAX!\n"; 1121 1122 } else { 1123 print "I'm not so sure about where $^O is...\n"; 1124 } 1125 1126On VMS, perl determines the UTC offset from the C<SYS$TIMEZONE_DIFFERENTIAL> 1127logical name. Although the VMS epoch began at 17-NOV-1858 00:00:00.00, 1128calls to C<localtime> are adjusted to count offsets from 112901-JAN-1970 00:00:00.00, just like Unix. 1130 1131Also see: 1132 1133=over 4 1134 1135=item * 1136 1137F<README.vms> (installed as L<README_vms>), L<perlvms> 1138 1139=item * 1140 1141vmsperl list, majordomo@perl.org 1142 1143(Put the words C<subscribe vmsperl> in message body.) 1144 1145=item * 1146 1147vmsperl on the web, http://www.sidhe.org/vmsperl/index.html 1148 1149=back 1150 1151=head2 VOS 1152 1153Perl on VOS is discussed in F<README.vos> in the perl distribution 1154(installed as L<perlvos>). Perl on VOS can accept either VOS- or 1155Unix-style file specifications as in either of the following: 1156 1157 C<< $ perl -ne "print if /perl_setup/i" >system>notices >> 1158 C<< $ perl -ne "print if /perl_setup/i" /system/notices >> 1159 1160or even a mixture of both as in: 1161 1162 C<< $ perl -ne "print if /perl_setup/i" >system/notices >> 1163 1164Even though VOS allows the slash character to appear in object 1165names, because the VOS port of Perl interprets it as a pathname 1166delimiting character, VOS files, directories, or links whose names 1167contain a slash character cannot be processed. Such files must be 1168renamed before they can be processed by Perl. Note that VOS limits 1169file names to 32 or fewer characters. 1170 1171Perl on VOS can be built using two different compilers and two different 1172versions of the POSIX runtime. The recommended method for building full 1173Perl is with the GNU C compiler and the generally-available version of 1174VOS POSIX support. See F<README.vos> (installed as L<perlvos>) for 1175restrictions that apply when Perl is built using the VOS Standard C 1176compiler or the alpha version of VOS POSIX support. 1177 1178The value of C<$^O> on VOS is "VOS". To determine the architecture that 1179you are running on without resorting to loading all of C<%Config> you 1180can examine the content of the @INC array like so: 1181 1182 if ($^O =~ /VOS/) { 1183 print "I'm on a Stratus box!\n"; 1184 } else { 1185 print "I'm not on a Stratus box!\n"; 1186 die; 1187 } 1188 1189 if (grep(/860/, @INC)) { 1190 print "This box is a Stratus XA/R!\n"; 1191 1192 } elsif (grep(/7100/, @INC)) { 1193 print "This box is a Stratus HP 7100 or 8xxx!\n"; 1194 1195 } elsif (grep(/8000/, @INC)) { 1196 print "This box is a Stratus HP 8xxx!\n"; 1197 1198 } else { 1199 print "This box is a Stratus 68K!\n"; 1200 } 1201 1202Also see: 1203 1204=over 4 1205 1206=item * 1207 1208F<README.vos> (installed as L<perlvos>) 1209 1210=item * 1211 1212The VOS mailing list. 1213 1214There is no specific mailing list for Perl on VOS. You can post 1215comments to the comp.sys.stratus newsgroup, or subscribe to the general 1216Stratus mailing list. Send a letter with "subscribe Info-Stratus" in 1217the message body to majordomo@list.stratagy.com. 1218 1219=item * 1220 1221VOS Perl on the web at http://ftp.stratus.com/pub/vos/posix/posix.html 1222 1223=back 1224 1225=head2 EBCDIC Platforms 1226 1227Recent versions of Perl have been ported to platforms such as OS/400 on 1228AS/400 minicomputers as well as OS/390, VM/ESA, and BS2000 for S/390 1229Mainframes. Such computers use EBCDIC character sets internally (usually 1230Character Code Set ID 0037 for OS/400 and either 1047 or POSIX-BC for S/390 1231systems). On the mainframe perl currently works under the "Unix system 1232services for OS/390" (formerly known as OpenEdition), VM/ESA OpenEdition, or 1233the BS200 POSIX-BC system (BS2000 is supported in perl 5.6 and greater). 1234See L<perlos390> for details. Note that for OS/400 there is also a port of 1235Perl 5.8.1/5.9.0 or later to the PASE which is ASCII-based (as opposed to 1236ILE which is EBCDIC-based), see L<perlos400>. 1237 1238As of R2.5 of USS for OS/390 and Version 2.3 of VM/ESA these Unix 1239sub-systems do not support the C<#!> shebang trick for script invocation. 1240Hence, on OS/390 and VM/ESA perl scripts can be executed with a header 1241similar to the following simple script: 1242 1243 : # use perl 1244 eval 'exec /usr/local/bin/perl -S $0 ${1+"$@"}' 1245 if 0; 1246 #!/usr/local/bin/perl # just a comment really 1247 1248 print "Hello from perl!\n"; 1249 1250OS/390 will support the C<#!> shebang trick in release 2.8 and beyond. 1251Calls to C<system> and backticks can use POSIX shell syntax on all 1252S/390 systems. 1253 1254On the AS/400, if PERL5 is in your library list, you may need 1255to wrap your perl scripts in a CL procedure to invoke them like so: 1256 1257 BEGIN 1258 CALL PGM(PERL5/PERL) PARM('/QOpenSys/hello.pl') 1259 ENDPGM 1260 1261This will invoke the perl script F<hello.pl> in the root of the 1262QOpenSys file system. On the AS/400 calls to C<system> or backticks 1263must use CL syntax. 1264 1265On these platforms, bear in mind that the EBCDIC character set may have 1266an effect on what happens with some perl functions (such as C<chr>, 1267C<pack>, C<print>, C<printf>, C<ord>, C<sort>, C<sprintf>, C<unpack>), as 1268well as bit-fiddling with ASCII constants using operators like C<^>, C<&> 1269and C<|>, not to mention dealing with socket interfaces to ASCII computers 1270(see L<"Newlines">). 1271 1272Fortunately, most web servers for the mainframe will correctly 1273translate the C<\n> in the following statement to its ASCII equivalent 1274(C<\r> is the same under both Unix and OS/390 & VM/ESA): 1275 1276 print "Content-type: text/html\r\n\r\n"; 1277 1278The values of C<$^O> on some of these platforms includes: 1279 1280 uname $^O $Config{'archname'} 1281 -------------------------------------------- 1282 OS/390 os390 os390 1283 OS400 os400 os400 1284 POSIX-BC posix-bc BS2000-posix-bc 1285 VM/ESA vmesa vmesa 1286 1287Some simple tricks for determining if you are running on an EBCDIC 1288platform could include any of the following (perhaps all): 1289 1290 if ("\t" eq "\05") { print "EBCDIC may be spoken here!\n"; } 1291 1292 if (ord('A') == 193) { print "EBCDIC may be spoken here!\n"; } 1293 1294 if (chr(169) eq 'z') { print "EBCDIC may be spoken here!\n"; } 1295 1296One thing you may not want to rely on is the EBCDIC encoding 1297of punctuation characters since these may differ from code page to code 1298page (and once your module or script is rumoured to work with EBCDIC, 1299folks will want it to work with all EBCDIC character sets). 1300 1301Also see: 1302 1303=over 4 1304 1305=item * 1306 1307* 1308 1309L<perlos390>, F<README.os390>, F<perlbs2000>, F<README.vmesa>, 1310L<perlebcdic>. 1311 1312=item * 1313 1314The perl-mvs@perl.org list is for discussion of porting issues as well as 1315general usage issues for all EBCDIC Perls. Send a message body of 1316"subscribe perl-mvs" to majordomo@perl.org. 1317 1318=item * 1319 1320AS/400 Perl information at 1321http://as400.rochester.ibm.com/ 1322as well as on CPAN in the F<ports/> directory. 1323 1324=back 1325 1326=head2 Acorn RISC OS 1327 1328Because Acorns use ASCII with newlines (C<\n>) in text files as C<\012> like 1329Unix, and because Unix filename emulation is turned on by default, 1330most simple scripts will probably work "out of the box". The native 1331filesystem is modular, and individual filesystems are free to be 1332case-sensitive or insensitive, and are usually case-preserving. Some 1333native filesystems have name length limits, which file and directory 1334names are silently truncated to fit. Scripts should be aware that the 1335standard filesystem currently has a name length limit of B<10> 1336characters, with up to 77 items in a directory, but other filesystems 1337may not impose such limitations. 1338 1339Native filenames are of the form 1340 1341 Filesystem#Special_Field::DiskName.$.Directory.Directory.File 1342 1343where 1344 1345 Special_Field is not usually present, but may contain . and $ . 1346 Filesystem =~ m|[A-Za-z0-9_]| 1347 DsicName =~ m|[A-Za-z0-9_/]| 1348 $ represents the root directory 1349 . is the path separator 1350 @ is the current directory (per filesystem but machine global) 1351 ^ is the parent directory 1352 Directory and File =~ m|[^\0- "\.\$\%\&:\@\\^\|\177]+| 1353 1354The default filename translation is roughly C<tr|/.|./|;> 1355 1356Note that C<"ADFS::HardDisk.$.File" ne 'ADFS::HardDisk.$.File'> and that 1357the second stage of C<$> interpolation in regular expressions will fall 1358foul of the C<$.> if scripts are not careful. 1359 1360Logical paths specified by system variables containing comma-separated 1361search lists are also allowed; hence C<System:Modules> is a valid 1362filename, and the filesystem will prefix C<Modules> with each section of 1363C<System$Path> until a name is made that points to an object on disk. 1364Writing to a new file C<System:Modules> would be allowed only if 1365C<System$Path> contains a single item list. The filesystem will also 1366expand system variables in filenames if enclosed in angle brackets, so 1367C<< <System$Dir>.Modules >> would look for the file 1368S<C<$ENV{'System$Dir'} . 'Modules'>>. The obvious implication of this is 1369that B<fully qualified filenames can start with C<< <> >>> and should 1370be protected when C<open> is used for input. 1371 1372Because C<.> was in use as a directory separator and filenames could not 1373be assumed to be unique after 10 characters, Acorn implemented the C 1374compiler to strip the trailing C<.c> C<.h> C<.s> and C<.o> suffix from 1375filenames specified in source code and store the respective files in 1376subdirectories named after the suffix. Hence files are translated: 1377 1378 foo.h h.foo 1379 C:foo.h C:h.foo (logical path variable) 1380 sys/os.h sys.h.os (C compiler groks Unix-speak) 1381 10charname.c c.10charname 1382 10charname.o o.10charname 1383 11charname_.c c.11charname (assuming filesystem truncates at 10) 1384 1385The Unix emulation library's translation of filenames to native assumes 1386that this sort of translation is required, and it allows a user-defined list 1387of known suffixes that it will transpose in this fashion. This may 1388seem transparent, but consider that with these rules C<foo/bar/baz.h> 1389and C<foo/bar/h/baz> both map to C<foo.bar.h.baz>, and that C<readdir> and 1390C<glob> cannot and do not attempt to emulate the reverse mapping. Other 1391C<.>'s in filenames are translated to C</>. 1392 1393As implied above, the environment accessed through C<%ENV> is global, and 1394the convention is that program specific environment variables are of the 1395form C<Program$Name>. Each filesystem maintains a current directory, 1396and the current filesystem's current directory is the B<global> current 1397directory. Consequently, sociable programs don't change the current 1398directory but rely on full pathnames, and programs (and Makefiles) cannot 1399assume that they can spawn a child process which can change the current 1400directory without affecting its parent (and everyone else for that 1401matter). 1402 1403Because native operating system filehandles are global and are currently 1404allocated down from 255, with 0 being a reserved value, the Unix emulation 1405library emulates Unix filehandles. Consequently, you can't rely on 1406passing C<STDIN>, C<STDOUT>, or C<STDERR> to your children. 1407 1408The desire of users to express filenames of the form 1409C<< <Foo$Dir>.Bar >> on the command line unquoted causes problems, 1410too: C<``> command output capture has to perform a guessing game. It 1411assumes that a string C<< <[^<>]+\$[^<>]> >> is a 1412reference to an environment variable, whereas anything else involving 1413C<< < >> or C<< > >> is redirection, and generally manages to be 99% 1414right. Of course, the problem remains that scripts cannot rely on any 1415Unix tools being available, or that any tools found have Unix-like command 1416line arguments. 1417 1418Extensions and XS are, in theory, buildable by anyone using free 1419tools. In practice, many don't, as users of the Acorn platform are 1420used to binary distributions. MakeMaker does run, but no available 1421make currently copes with MakeMaker's makefiles; even if and when 1422this should be fixed, the lack of a Unix-like shell will cause 1423problems with makefile rules, especially lines of the form C<cd 1424sdbm && make all>, and anything using quoting. 1425 1426"S<RISC OS>" is the proper name for the operating system, but the value 1427in C<$^O> is "riscos" (because we don't like shouting). 1428 1429=head2 Other perls 1430 1431Perl has been ported to many platforms that do not fit into any of 1432the categories listed above. Some, such as AmigaOS, Atari MiNT, 1433BeOS, HP MPE/iX, QNX, Plan 9, and VOS, have been well-integrated 1434into the standard Perl source code kit. You may need to see the 1435F<ports/> directory on CPAN for information, and possibly binaries, 1436for the likes of: aos, Atari ST, lynxos, riscos, Novell Netware, 1437Tandem Guardian, I<etc.> (Yes, we know that some of these OSes may 1438fall under the Unix category, but we are not a standards body.) 1439 1440Some approximate operating system names and their C<$^O> values 1441in the "OTHER" category include: 1442 1443 OS $^O $Config{'archname'} 1444 ------------------------------------------ 1445 Amiga DOS amigaos m68k-amigos 1446 BeOS beos 1447 MPE/iX mpeix PA-RISC1.1 1448 1449See also: 1450 1451=over 4 1452 1453=item * 1454 1455Amiga, F<README.amiga> (installed as L<perlamiga>). 1456 1457=item * 1458 1459Atari, F<README.mint> and Guido Flohr's web page 1460http://stud.uni-sb.de/~gufl0000/ 1461 1462=item * 1463 1464Be OS, F<README.beos> 1465 1466=item * 1467 1468HP 300 MPE/iX, F<README.mpeix> and Mark Bixby's web page 1469http://www.bixby.org/mark/perlix.html 1470 1471=item * 1472 1473A free perl5-based PERL.NLM for Novell Netware is available in 1474precompiled binary and source code form from http://www.novell.com/ 1475as well as from CPAN. 1476 1477=item * 1478 1479S<Plan 9>, F<README.plan9> 1480 1481=back 1482 1483=head1 FUNCTION IMPLEMENTATIONS 1484 1485Listed below are functions that are either completely unimplemented 1486or else have been implemented differently on various platforms. 1487Following each description will be, in parentheses, a list of 1488platforms that the description applies to. 1489 1490The list may well be incomplete, or even wrong in some places. When 1491in doubt, consult the platform-specific README files in the Perl 1492source distribution, and any other documentation resources accompanying 1493a given port. 1494 1495Be aware, moreover, that even among Unix-ish systems there are variations. 1496 1497For many functions, you can also query C<%Config>, exported by 1498default from the Config module. For example, to check whether the 1499platform has the C<lstat> call, check C<$Config{d_lstat}>. See 1500L<Config> for a full description of available variables. 1501 1502=head2 Alphabetical Listing of Perl Functions 1503 1504=over 8 1505 1506=item -X FILEHANDLE 1507 1508=item -X EXPR 1509 1510=item -X 1511 1512C<-r>, C<-w>, and C<-x> have a limited meaning only; directories 1513and applications are executable, and there are no uid/gid 1514considerations. C<-o> is not supported. (S<Mac OS>) 1515 1516C<-r>, C<-w>, C<-x>, and C<-o> tell whether the file is accessible, 1517which may not reflect UIC-based file protections. (VMS) 1518 1519C<-s> returns the size of the data fork, not the total size of data fork 1520plus resource fork. (S<Mac OS>). 1521 1522C<-s> by name on an open file will return the space reserved on disk, 1523rather than the current extent. C<-s> on an open filehandle returns the 1524current size. (S<RISC OS>) 1525 1526C<-R>, C<-W>, C<-X>, C<-O> are indistinguishable from C<-r>, C<-w>, 1527C<-x>, C<-o>. (S<Mac OS>, Win32, VMS, S<RISC OS>) 1528 1529C<-b>, C<-c>, C<-k>, C<-g>, C<-p>, C<-u>, C<-A> are not implemented. 1530(S<Mac OS>) 1531 1532C<-g>, C<-k>, C<-l>, C<-p>, C<-u>, C<-A> are not particularly meaningful. 1533(Win32, VMS, S<RISC OS>) 1534 1535C<-d> is true if passed a device spec without an explicit directory. 1536(VMS) 1537 1538C<-T> and C<-B> are implemented, but might misclassify Mac text files 1539with foreign characters; this is the case will all platforms, but may 1540affect S<Mac OS> often. (S<Mac OS>) 1541 1542C<-x> (or C<-X>) determine if a file ends in one of the executable 1543suffixes. C<-S> is meaningless. (Win32) 1544 1545C<-x> (or C<-X>) determine if a file has an executable file type. 1546(S<RISC OS>) 1547 1548=item binmode FILEHANDLE 1549 1550Meaningless. (S<Mac OS>, S<RISC OS>) 1551 1552Reopens file and restores pointer; if function fails, underlying 1553filehandle may be closed, or pointer may be in a different position. 1554(VMS) 1555 1556The value returned by C<tell> may be affected after the call, and 1557the filehandle may be flushed. (Win32) 1558 1559=item chmod LIST 1560 1561Only limited meaning. Disabling/enabling write permission is mapped to 1562locking/unlocking the file. (S<Mac OS>) 1563 1564Only good for changing "owner" read-write access, "group", and "other" 1565bits are meaningless. (Win32) 1566 1567Only good for changing "owner" and "other" read-write access. (S<RISC OS>) 1568 1569Access permissions are mapped onto VOS access-control list changes. (VOS) 1570 1571The actual permissions set depend on the value of the C<CYGWIN> 1572in the SYSTEM environment settings. (Cygwin) 1573 1574=item chown LIST 1575 1576Not implemented. (S<Mac OS>, Win32, S<Plan 9>, S<RISC OS>, VOS) 1577 1578Does nothing, but won't fail. (Win32) 1579 1580=item chroot FILENAME 1581 1582=item chroot 1583 1584Not implemented. (S<Mac OS>, Win32, VMS, S<Plan 9>, S<RISC OS>, VOS, VM/ESA) 1585 1586=item crypt PLAINTEXT,SALT 1587 1588May not be available if library or source was not provided when building 1589perl. (Win32) 1590 1591Not implemented. (VOS) 1592 1593=item dbmclose HASH 1594 1595Not implemented. (VMS, S<Plan 9>, VOS) 1596 1597=item dbmopen HASH,DBNAME,MODE 1598 1599Not implemented. (VMS, S<Plan 9>, VOS) 1600 1601=item dump LABEL 1602 1603Not useful. (S<Mac OS>, S<RISC OS>) 1604 1605Not implemented. (Win32) 1606 1607Invokes VMS debugger. (VMS) 1608 1609=item exec LIST 1610 1611Not implemented. (S<Mac OS>) 1612 1613Implemented via Spawn. (VM/ESA) 1614 1615Does not automatically flush output handles on some platforms. 1616(SunOS, Solaris, HP-UX) 1617 1618=item exit EXPR 1619 1620=item exit 1621 1622Emulates UNIX exit() (which considers C<exit 1> to indicate an error) by 1623mapping the C<1> to SS$_ABORT (C<44>). This behavior may be overridden 1624with the pragma C<use vmsish 'exit'>. As with the CRTL's exit() 1625function, C<exit 0> is also mapped to an exit status of SS$_NORMAL 1626(C<1>); this mapping cannot be overridden. Any other argument to exit() 1627is used directly as Perl's exit status. (VMS) 1628 1629=item fcntl FILEHANDLE,FUNCTION,SCALAR 1630 1631Not implemented. (Win32, VMS) 1632 1633=item flock FILEHANDLE,OPERATION 1634 1635Not implemented (S<Mac OS>, VMS, S<RISC OS>, VOS). 1636 1637Available only on Windows NT (not on Windows 95). (Win32) 1638 1639=item fork 1640 1641Not implemented. (S<Mac OS>, AmigaOS, S<RISC OS>, VOS, VM/ESA, VMS) 1642 1643Emulated using multiple interpreters. See L<perlfork>. (Win32) 1644 1645Does not automatically flush output handles on some platforms. 1646(SunOS, Solaris, HP-UX) 1647 1648=item getlogin 1649 1650Not implemented. (S<Mac OS>, S<RISC OS>) 1651 1652=item getpgrp PID 1653 1654Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS) 1655 1656=item getppid 1657 1658Not implemented. (S<Mac OS>, Win32, S<RISC OS>) 1659 1660=item getpriority WHICH,WHO 1661 1662Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS, VM/ESA) 1663 1664=item getpwnam NAME 1665 1666Not implemented. (S<Mac OS>, Win32) 1667 1668Not useful. (S<RISC OS>) 1669 1670=item getgrnam NAME 1671 1672Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>) 1673 1674=item getnetbyname NAME 1675 1676Not implemented. (S<Mac OS>, Win32, S<Plan 9>) 1677 1678=item getpwuid UID 1679 1680Not implemented. (S<Mac OS>, Win32) 1681 1682Not useful. (S<RISC OS>) 1683 1684=item getgrgid GID 1685 1686Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>) 1687 1688=item getnetbyaddr ADDR,ADDRTYPE 1689 1690Not implemented. (S<Mac OS>, Win32, S<Plan 9>) 1691 1692=item getprotobynumber NUMBER 1693 1694Not implemented. (S<Mac OS>) 1695 1696=item getservbyport PORT,PROTO 1697 1698Not implemented. (S<Mac OS>) 1699 1700=item getpwent 1701 1702Not implemented. (S<Mac OS>, Win32, VM/ESA) 1703 1704=item getgrent 1705 1706Not implemented. (S<Mac OS>, Win32, VMS, VM/ESA) 1707 1708=item gethostbyname 1709 1710C<gethostbyname('localhost')> does not work everywhere: you may have 1711to use C<gethostbyname('127.0.0.1')>. (S<Mac OS>, S<Irix 5>) 1712 1713=item gethostent 1714 1715Not implemented. (S<Mac OS>, Win32) 1716 1717=item getnetent 1718 1719Not implemented. (S<Mac OS>, Win32, S<Plan 9>) 1720 1721=item getprotoent 1722 1723Not implemented. (S<Mac OS>, Win32, S<Plan 9>) 1724 1725=item getservent 1726 1727Not implemented. (Win32, S<Plan 9>) 1728 1729=item sethostent STAYOPEN 1730 1731Not implemented. (S<Mac OS>, Win32, S<Plan 9>, S<RISC OS>) 1732 1733=item setnetent STAYOPEN 1734 1735Not implemented. (S<Mac OS>, Win32, S<Plan 9>, S<RISC OS>) 1736 1737=item setprotoent STAYOPEN 1738 1739Not implemented. (S<Mac OS>, Win32, S<Plan 9>, S<RISC OS>) 1740 1741=item setservent STAYOPEN 1742 1743Not implemented. (S<Plan 9>, Win32, S<RISC OS>) 1744 1745=item endpwent 1746 1747Not implemented. (S<Mac OS>, MPE/iX, VM/ESA, Win32) 1748 1749=item endgrent 1750 1751Not implemented. (S<Mac OS>, MPE/iX, S<RISC OS>, VM/ESA, VMS, Win32) 1752 1753=item endhostent 1754 1755Not implemented. (S<Mac OS>, Win32) 1756 1757=item endnetent 1758 1759Not implemented. (S<Mac OS>, Win32, S<Plan 9>) 1760 1761=item endprotoent 1762 1763Not implemented. (S<Mac OS>, Win32, S<Plan 9>) 1764 1765=item endservent 1766 1767Not implemented. (S<Plan 9>, Win32) 1768 1769=item getsockopt SOCKET,LEVEL,OPTNAME 1770 1771Not implemented. (S<Plan 9>) 1772 1773=item glob EXPR 1774 1775=item glob 1776 1777This operator is implemented via the File::Glob extension on most 1778platforms. See L<File::Glob> for portability information. 1779 1780=item ioctl FILEHANDLE,FUNCTION,SCALAR 1781 1782Not implemented. (VMS) 1783 1784Available only for socket handles, and it does what the ioctlsocket() call 1785in the Winsock API does. (Win32) 1786 1787Available only for socket handles. (S<RISC OS>) 1788 1789=item kill SIGNAL, LIST 1790 1791C<kill(0, LIST)> is implemented for the sake of taint checking; 1792use with other signals is unimplemented. (S<Mac OS>) 1793 1794Not implemented, hence not useful for taint checking. (S<RISC OS>) 1795 1796C<kill()> doesn't have the semantics of C<raise()>, i.e. it doesn't send 1797a signal to the identified process like it does on Unix platforms. 1798Instead C<kill($sig, $pid)> terminates the process identified by $pid, 1799and makes it exit immediately with exit status $sig. As in Unix, if 1800$sig is 0 and the specified process exists, it returns true without 1801actually terminating it. (Win32) 1802 1803=item link OLDFILE,NEWFILE 1804 1805Not implemented. (S<Mac OS>, MPE/iX, VMS, S<RISC OS>) 1806 1807Link count not updated because hard links are not quite that hard 1808(They are sort of half-way between hard and soft links). (AmigaOS) 1809 1810Hard links are implemented on Win32 (Windows NT and Windows 2000) 1811under NTFS only. 1812 1813=item lstat FILEHANDLE 1814 1815=item lstat EXPR 1816 1817=item lstat 1818 1819Not implemented. (VMS, S<RISC OS>) 1820 1821Return values (especially for device and inode) may be bogus. (Win32) 1822 1823=item msgctl ID,CMD,ARG 1824 1825=item msgget KEY,FLAGS 1826 1827=item msgsnd ID,MSG,FLAGS 1828 1829=item msgrcv ID,VAR,SIZE,TYPE,FLAGS 1830 1831Not implemented. (S<Mac OS>, Win32, VMS, S<Plan 9>, S<RISC OS>, VOS) 1832 1833=item open FILEHANDLE,EXPR 1834 1835=item open FILEHANDLE 1836 1837The C<|> variants are supported only if ToolServer is installed. 1838(S<Mac OS>) 1839 1840open to C<|-> and C<-|> are unsupported. (S<Mac OS>, Win32, S<RISC OS>) 1841 1842Opening a process does not automatically flush output handles on some 1843platforms. (SunOS, Solaris, HP-UX) 1844 1845=item pipe READHANDLE,WRITEHANDLE 1846 1847Very limited functionality. (MiNT) 1848 1849=item readlink EXPR 1850 1851=item readlink 1852 1853Not implemented. (Win32, VMS, S<RISC OS>) 1854 1855=item select RBITS,WBITS,EBITS,TIMEOUT 1856 1857Only implemented on sockets. (Win32, VMS) 1858 1859Only reliable on sockets. (S<RISC OS>) 1860 1861Note that the C<select FILEHANDLE> form is generally portable. 1862 1863=item semctl ID,SEMNUM,CMD,ARG 1864 1865=item semget KEY,NSEMS,FLAGS 1866 1867=item semop KEY,OPSTRING 1868 1869Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS) 1870 1871=item setgrent 1872 1873Not implemented. (S<Mac OS>, MPE/iX, VMS, Win32, S<RISC OS>) 1874 1875=item setpgrp PID,PGRP 1876 1877Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS) 1878 1879=item setpriority WHICH,WHO,PRIORITY 1880 1881Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS) 1882 1883=item setpwent 1884 1885Not implemented. (S<Mac OS>, MPE/iX, Win32, S<RISC OS>) 1886 1887=item setsockopt SOCKET,LEVEL,OPTNAME,OPTVAL 1888 1889Not implemented. (S<Plan 9>) 1890 1891=item shmctl ID,CMD,ARG 1892 1893=item shmget KEY,SIZE,FLAGS 1894 1895=item shmread ID,VAR,POS,SIZE 1896 1897=item shmwrite ID,STRING,POS,SIZE 1898 1899Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS) 1900 1901=item sockatmark SOCKET 1902 1903A relatively recent addition to socket functions, may not 1904be implemented even in UNIX platforms. 1905 1906=item socketpair SOCKET1,SOCKET2,DOMAIN,TYPE,PROTOCOL 1907 1908Not implemented. (Win32, VMS, S<RISC OS>, VOS, VM/ESA) 1909 1910=item stat FILEHANDLE 1911 1912=item stat EXPR 1913 1914=item stat 1915 1916Platforms that do not have rdev, blksize, or blocks will return these 1917as '', so numeric comparison or manipulation of these fields may cause 1918'not numeric' warnings. 1919 1920mtime and atime are the same thing, and ctime is creation time instead of 1921inode change time. (S<Mac OS>). 1922 1923ctime not supported on UFS (S<Mac OS X>). 1924 1925ctime is creation time instead of inode change time (Win32). 1926 1927device and inode are not meaningful. (Win32) 1928 1929device and inode are not necessarily reliable. (VMS) 1930 1931mtime, atime and ctime all return the last modification time. Device and 1932inode are not necessarily reliable. (S<RISC OS>) 1933 1934dev, rdev, blksize, and blocks are not available. inode is not 1935meaningful and will differ between stat calls on the same file. (os2) 1936 1937some versions of cygwin when doing a stat("foo") and if not finding it 1938may then attempt to stat("foo.exe") (Cygwin) 1939 1940=item symlink OLDFILE,NEWFILE 1941 1942Not implemented. (Win32, VMS, S<RISC OS>) 1943 1944=item syscall LIST 1945 1946Not implemented. (S<Mac OS>, Win32, VMS, S<RISC OS>, VOS, VM/ESA) 1947 1948=item sysopen FILEHANDLE,FILENAME,MODE,PERMS 1949 1950The traditional "0", "1", and "2" MODEs are implemented with different 1951numeric values on some systems. The flags exported by C<Fcntl> 1952(O_RDONLY, O_WRONLY, O_RDWR) should work everywhere though. (S<Mac 1953OS>, OS/390, VM/ESA) 1954 1955=item system LIST 1956 1957In general, do not assume the UNIX/POSIX semantics that you can shift 1958C<$?> right by eight to get the exit value, or that C<$? & 127> 1959would give you the number of the signal that terminated the program, 1960or that C<$? & 128> would test true if the program was terminated by a 1961coredump. Instead, use the POSIX W*() interfaces: for example, use 1962WIFEXITED($?) and WEXITVALUE($?) to test for a normal exit and the exit 1963value, WIFSIGNALED($?) and WTERMSIG($?) for a signal exit and the 1964signal. Core dumping is not a portable concept, so there's no portable 1965way to test for that. 1966 1967Only implemented if ToolServer is installed. (S<Mac OS>) 1968 1969As an optimization, may not call the command shell specified in 1970C<$ENV{PERL5SHELL}>. C<system(1, @args)> spawns an external 1971process and immediately returns its process designator, without 1972waiting for it to terminate. Return value may be used subsequently 1973in C<wait> or C<waitpid>. Failure to spawn() a subprocess is indicated 1974by setting $? to "255 << 8". C<$?> is set in a way compatible with 1975Unix (i.e. the exitstatus of the subprocess is obtained by "$? >> 8", 1976as described in the documentation). (Win32) 1977 1978There is no shell to process metacharacters, and the native standard is 1979to pass a command line terminated by "\n" "\r" or "\0" to the spawned 1980program. Redirection such as C<< > foo >> is performed (if at all) by 1981the run time library of the spawned program. C<system> I<list> will call 1982the Unix emulation library's C<exec> emulation, which attempts to provide 1983emulation of the stdin, stdout, stderr in force in the parent, providing 1984the child program uses a compatible version of the emulation library. 1985I<scalar> will call the native command line direct and no such emulation 1986of a child Unix program will exists. Mileage B<will> vary. (S<RISC OS>) 1987 1988Far from being POSIX compliant. Because there may be no underlying 1989/bin/sh tries to work around the problem by forking and execing the 1990first token in its argument string. Handles basic redirection 1991("<" or ">") on its own behalf. (MiNT) 1992 1993Does not automatically flush output handles on some platforms. 1994(SunOS, Solaris, HP-UX) 1995 1996The return value is POSIX-like (shifted up by 8 bits), which only allows 1997room for a made-up value derived from the severity bits of the native 199832-bit condition code (unless overridden by C<use vmsish 'status'>). 1999For more details see L<perlvms/$?>. (VMS) 2000 2001=item times 2002 2003Only the first entry returned is nonzero. (S<Mac OS>) 2004 2005"cumulative" times will be bogus. On anything other than Windows NT 2006or Windows 2000, "system" time will be bogus, and "user" time is 2007actually the time returned by the clock() function in the C runtime 2008library. (Win32) 2009 2010Not useful. (S<RISC OS>) 2011 2012=item truncate FILEHANDLE,LENGTH 2013 2014=item truncate EXPR,LENGTH 2015 2016Not implemented. (Older versions of VMS) 2017 2018Truncation to zero-length only. (VOS) 2019 2020If a FILEHANDLE is supplied, it must be writable and opened in append 2021mode (i.e., use C<<< open(FH, '>>filename') >>> 2022or C<sysopen(FH,...,O_APPEND|O_RDWR)>. If a filename is supplied, it 2023should not be held open elsewhere. (Win32) 2024 2025=item umask EXPR 2026 2027=item umask 2028 2029Returns undef where unavailable, as of version 5.005. 2030 2031C<umask> works but the correct permissions are set only when the file 2032is finally closed. (AmigaOS) 2033 2034=item utime LIST 2035 2036Only the modification time is updated. (S<BeOS>, S<Mac OS>, VMS, S<RISC OS>) 2037 2038May not behave as expected. Behavior depends on the C runtime 2039library's implementation of utime(), and the filesystem being 2040used. The FAT filesystem typically does not support an "access 2041time" field, and it may limit timestamps to a granularity of 2042two seconds. (Win32) 2043 2044=item wait 2045 2046=item waitpid PID,FLAGS 2047 2048Not implemented. (S<Mac OS>, VOS) 2049 2050Can only be applied to process handles returned for processes spawned 2051using C<system(1, ...)> or pseudo processes created with C<fork()>. (Win32) 2052 2053Not useful. (S<RISC OS>) 2054 2055=back 2056 2057=head1 CHANGES 2058 2059=over 4 2060 2061=item v1.48, 02 February 2001 2062 2063Various updates from perl5-porters over the past year, supported 2064platforms update from Jarkko Hietaniemi. 2065 2066=item v1.47, 22 March 2000 2067 2068Various cleanups from Tom Christiansen, including migration of 2069long platform listings from L<perl>. 2070 2071=item v1.46, 12 February 2000 2072 2073Updates for VOS and MPE/iX. (Peter Prymmer) Other small changes. 2074 2075=item v1.45, 20 December 1999 2076 2077Small changes from 5.005_63 distribution, more changes to EBCDIC info. 2078 2079=item v1.44, 19 July 1999 2080 2081A bunch of updates from Peter Prymmer for C<$^O> values, 2082endianness, File::Spec, VMS, BS2000, OS/400. 2083 2084=item v1.43, 24 May 1999 2085 2086Added a lot of cleaning up from Tom Christiansen. 2087 2088=item v1.42, 22 May 1999 2089 2090Added notes about tests, sprintf/printf, and epoch offsets. 2091 2092=item v1.41, 19 May 1999 2093 2094Lots more little changes to formatting and content. 2095 2096Added a bunch of C<$^O> and related values 2097for various platforms; fixed mail and web addresses, and added 2098and changed miscellaneous notes. (Peter Prymmer) 2099 2100=item v1.40, 11 April 1999 2101 2102Miscellaneous changes. 2103 2104=item v1.39, 11 February 1999 2105 2106Changes from Jarkko and EMX URL fixes Michael Schwern. Additional 2107note about newlines added. 2108 2109=item v1.38, 31 December 1998 2110 2111More changes from Jarkko. 2112 2113=item v1.37, 19 December 1998 2114 2115More minor changes. Merge two separate version 1.35 documents. 2116 2117=item v1.36, 9 September 1998 2118 2119Updated for Stratus VOS. Also known as version 1.35. 2120 2121=item v1.35, 13 August 1998 2122 2123Integrate more minor changes, plus addition of new sections under 2124L<"ISSUES">: L<"Numbers endianness and Width">, 2125L<"Character sets and character encoding">, 2126L<"Internationalisation">. 2127 2128=item v1.33, 06 August 1998 2129 2130Integrate more minor changes. 2131 2132=item v1.32, 05 August 1998 2133 2134Integrate more minor changes. 2135 2136=item v1.30, 03 August 1998 2137 2138Major update for RISC OS, other minor changes. 2139 2140=item v1.23, 10 July 1998 2141 2142First public release with perl5.005. 2143 2144=back 2145 2146=head1 Supported Platforms 2147 2148As of September 2003 (the Perl release 5.8.1), the following platforms 2149are able to build Perl from the standard source code distribution 2150available at http://www.cpan.org/src/index.html 2151 2152 AIX 2153 BeOS 2154 BSD/OS (BSDi) 2155 Cygwin 2156 DG/UX 2157 DOS DJGPP 1) 2158 DYNIX/ptx 2159 EPOC R5 2160 FreeBSD 2161 HI-UXMPP (Hitachi) (5.8.0 worked but we didn't know it) 2162 HP-UX 2163 IRIX 2164 Linux 2165 LynxOS 2166 Mac OS Classic 2167 Mac OS X (Darwin) 2168 MPE/iX 2169 NetBSD 2170 NetWare 2171 NonStop-UX 2172 ReliantUNIX (formerly SINIX) 2173 OpenBSD 2174 OpenVMS (formerly VMS) 2175 Open UNIX (Unixware) (since Perl 5.8.1/5.9.0) 2176 OS/2 2177 OS/400 (using the PASE) (since Perl 5.8.1/5.9.0) 2178 PowerUX 2179 POSIX-BC (formerly BS2000) 2180 QNX 2181 Solaris 2182 SunOS 4 2183 SUPER-UX (NEC) 2184 SVR4 2185 Tru64 UNIX (formerly DEC OSF/1, Digital UNIX) 2186 UNICOS 2187 UNICOS/mk 2188 UTS 2189 VOS 2190 Win95/98/ME/2K/XP 2) 2191 WinCE 2192 z/OS (formerly OS/390) 2193 VM/ESA 2194 2195 1) in DOS mode either the DOS or OS/2 ports can be used 2196 2) compilers: Borland, MinGW (GCC), VC6 2197 2198The following platforms worked with the previous releases (5.6 and 21995.7), but we did not manage either to fix or to test these in time 2200for the 5.8.1 release. There is a very good chance that many of these 2201will work fine with the 5.8.1. 2202 2203 DomainOS 2204 Hurd 2205 MachTen 2206 PowerMAX 2207 SCO SV 2208 Unixware 2209 Windows 3.1 2210 2211Known to be broken for 5.8.0 and 5.8.1 (but 5.6.1 and 5.7.2 can be used): 2212 2213 AmigaOS 2214 2215The following platforms have been known to build Perl from source in 2216the past (5.005_03 and earlier), but we haven't been able to verify 2217their status for the current release, either because the 2218hardware/software platforms are rare or because we don't have an 2219active champion on these platforms--or both. They used to work, 2220though, so go ahead and try compiling them, and let perlbug@perl.org 2221of any trouble. 2222 2223 3b1 2224 A/UX 2225 ConvexOS 2226 CX/UX 2227 DC/OSx 2228 DDE SMES 2229 DOS EMX 2230 Dynix 2231 EP/IX 2232 ESIX 2233 FPS 2234 GENIX 2235 Greenhills 2236 ISC 2237 MachTen 68k 2238 MiNT 2239 MPC 2240 NEWS-OS 2241 NextSTEP 2242 OpenSTEP 2243 Opus 2244 Plan 9 2245 RISC/os 2246 SCO ODT/OSR 2247 Stellar 2248 SVR2 2249 TI1500 2250 TitanOS 2251 Ultrix 2252 Unisys Dynix 2253 2254The following platforms have their own source code distributions and 2255binaries available via http://www.cpan.org/ports/ 2256 2257 Perl release 2258 2259 OS/400 (ILE) 5.005_02 2260 Tandem Guardian 5.004 2261 2262The following platforms have only binaries available via 2263http://www.cpan.org/ports/index.html : 2264 2265 Perl release 2266 2267 Acorn RISCOS 5.005_02 2268 AOS 5.002 2269 LynxOS 5.004_02 2270 2271Although we do suggest that you always build your own Perl from 2272the source code, both for maximal configurability and for security, 2273in case you are in a hurry you can check 2274http://www.cpan.org/ports/index.html for binary distributions. 2275 2276=head1 SEE ALSO 2277 2278L<perlaix>, L<perlamiga>, L<perlapollo>, L<perlbeos>, L<perlbs2000>, 2279L<perlce>, L<perlcygwin>, L<perldgux>, L<perldos>, L<perlepoc>, 2280L<perlebcdic>, L<perlfreebsd>, L<perlhurd>, L<perlhpux>, L<perlirix>, 2281L<perlmachten>, L<perlmacos>, L<perlmacosx>, L<perlmint>, L<perlmpeix>, 2282L<perlnetware>, L<perlos2>, L<perlos390>, L<perlos400>, 2283L<perlplan9>, L<perlqnx>, L<perlsolaris>, L<perltru64>, 2284L<perlunicode>, L<perlvmesa>, L<perlvms>, L<perlvos>, 2285L<perlwin32>, and L<Win32>. 2286 2287=head1 AUTHORS / CONTRIBUTORS 2288 2289Abigail <abigail@foad.org>, 2290Charles Bailey <bailey@newman.upenn.edu>, 2291Graham Barr <gbarr@pobox.com>, 2292Tom Christiansen <tchrist@perl.com>, 2293Nicholas Clark <nick@ccl4.org>, 2294Thomas Dorner <Thomas.Dorner@start.de>, 2295Andy Dougherty <doughera@lafayette.edu>, 2296Dominic Dunlop <domo@computer.org>, 2297Neale Ferguson <neale@vma.tabnsw.com.au>, 2298David J. Fiander <davidf@mks.com>, 2299Paul Green <Paul_Green@stratus.com>, 2300M.J.T. Guy <mjtg@cam.ac.uk>, 2301Jarkko Hietaniemi <jhi@iki.fi>, 2302Luther Huffman <lutherh@stratcom.com>, 2303Nick Ing-Simmons <nick@ing-simmons.net>, 2304Andreas J. KE<ouml>nig <a.koenig@mind.de>, 2305Markus Laker <mlaker@contax.co.uk>, 2306Andrew M. Langmead <aml@world.std.com>, 2307Larry Moore <ljmoore@freespace.net>, 2308Paul Moore <Paul.Moore@uk.origin-it.com>, 2309Chris Nandor <pudge@pobox.com>, 2310Matthias Neeracher <neeracher@mac.com>, 2311Philip Newton <pne@cpan.org>, 2312Gary Ng <71564.1743@CompuServe.COM>, 2313Tom Phoenix <rootbeer@teleport.com>, 2314AndrE<eacute> Pirard <A.Pirard@ulg.ac.be>, 2315Peter Prymmer <pvhp@forte.com>, 2316Hugo van der Sanden <hv@crypt0.demon.co.uk>, 2317Gurusamy Sarathy <gsar@activestate.com>, 2318Paul J. Schinder <schinder@pobox.com>, 2319Michael G Schwern <schwern@pobox.com>, 2320Dan Sugalski <dan@sidhe.org>, 2321Nathan Torkington <gnat@frii.com>. 2322 2323
