This document is meant to help you to find out what constitutes portable Perl code, so that once you have made your decision to write portably, you know where the lines are drawn, and you can stay within them.
There is a tradeoff between taking full advantage of a particular type of computer, and taking advantage of a full range of them. Naturally, as you make your range bigger (and thus more diverse), the common denominators drop, and you are left with fewer areas of common ground in which you can operate to accomplish a particular task. Thus, when you begin attacking a problem, it is important to consider which part of the tradeoff curve you want to operate under. Specifically, whether it is important to you that the task that you are coding needs the full generality of being portable, or if it is sufficient to just get the job done. This is the hardest choice to be made. The rest is easy, because Perl provides lots of choices, whichever way you want to approach your problem.
Looking at it another way, writing portable code is usually about willfully limiting your available choices. Naturally, it takes discipline to do that.
Be aware of two important points:
Here's the general rule: When you approach a task that is commonly done using a whole range of platforms, think in terms of writing portable code. That way, you don't sacrifice much by way of the implementation choices you can avail yourself of, and at the same time you can give your users lots of platform choices. On the other hand, when you have to take advantage of some unique feature of a particular platform, as is often the case with systems programming (whether for Unix, Windows, Mac OS, VMS, etc.), consider writing platform-specific code.
When the code will run on only two or three operating systems, then you may only need to consider the differences of those particular systems. The important thing is to decide where the code will run, and to be deliberate in your decision.
The material below is separated into three main sections: main issues of portability (the section on ISSUES, platform-specific issues (the section on PLATFORMS, and builtin perl functions that behave differently on various ports (the section on FUNCTION IMPLEMENTATIONS.
This information should not be considered complete; it includes possibly transient information about idiosyncrasies of some of the ports, almost all of which are in a state of constant evolution. Thus this material should be considered a perpetual work in progress (<IMG SRC="yellow_sign.gif'' ALT="Under Construction">).
Perl uses \n to represent the ``logical'' newline, where what is logical may depend on the platform in use. In MacPerl, \n always means \015. In DOSish perls, \n usually means \012, but when accessing a file in ``text'' mode, STDIO translates it to (or from) \015\012.
Due to the ``text'' mode translation, DOSish perls have limitations of using seek and tell when a file is being accessed in ``text'' mode. Specifically, if you stick to seek-ing to locations you got from tell (and no others), you are usually free to use seek and tell even in ``text'' mode. In general, using seek or tell or other file operations that count bytes instead of characters, without considering the length of \n, may be non-portable. If you use binmode on a file, however, you can usually use seek and tell with arbitrary values quite safely.
A common misconception in socket programming is that \n eq \012 everywhere. When using protocols such as common Internet protocols, \012 and \015 are called for specifically, and the values of the logical \n and \r (carriage return) are not reliable.
print SOCKET "Hi there, client!\r\n"; # WRONG print SOCKET "Hi there, client!\015\012"; # RIGHT[NOTE: this does not necessarily apply to communications that are filtered by another program or module before sending to the socket; the the most popular EBCDIC webserver, for instance, accepts \r\n, which translates those characters, along with all other characters in text streams, from EBCDIC to ASCII.]
However, using \015\012 (or \cM\cJ, or \x0D\x0A) can be tedious and unsightly, as well as confusing to those maintaining the code. As such, the Socket module supplies the Right Thing for those who want it.
use Socket qw(:DEFAULT :crlf); print SOCKET "Hi there, client!$CRLF" # RIGHTWhen reading from a socket, remember that the default input record separator ($/) is \n, but code like this should recognize $/ as \012 or \015\012:
while (<SOCKET>) { # ... }Better:
use Socket qw(:DEFAULT :crlf); local($/) = LF; # not needed if $/ is already \012
while (<SOCKET>) { s/$CR?$LF/\n/; # not sure if socket uses LF or CRLF, OK # s/\015?\012/\n/; # same thing }And this example is actually better than the previous one even for Unix platforms, because now any \015's (\cM's) are stripped out (and there was much rejoicing).
Conflicting storage orders make utter mess out of the numbers: if a little-endian host (Intel, Alpha) stores 0x12345678 (305419896 in decimal), a big-endian host (Motorola, MIPS, Sparc, PA) reads it as 0x78563412 (2018915346 in decimal). To avoid this problem in network (socket) connections use the pack() and unpack() formats "n" and "N", the ``network'' orders, they are guaranteed to be portable.
Different widths can cause truncation even between platforms of equal endianness: the platform of shorter width loses the upper parts of the number. There is no good solution for this problem except to avoid transferring or storing raw binary numbers.
One can circumnavigate both these problems in two ways: either transfer and store numbers always in text format, instead of raw binary, or consider using modules like Data::Dumper (included in the standard distribution as of Perl 5.005) and Storable.
While they are similar, file path specifications differ between Unix, Windows, Mac OS, OS/2, VMS, RISC OS and probably others. Unix, for example, is one of the few OSes that has the idea of a single root directory.
VMS, Windows, and OS/2 can work similarly to Unix with / as path separator, or in their own idiosyncratic ways (such as having several root directories and various ``unrooted'' device files such NIL: and LPT:).
Mac OS uses : as a path separator instead of /.
RISC OS perl can emulate Unix filenames with / as path separator, or go native and use . for path separator and : to signal filing systems and disc names.
As with the newline problem above, there are modules that can help. The File::Spec modules provide methods to do the Right Thing on whatever platform happens to be running the program.
use File::Spec; chdir(File::Spec->updir()); # go up one directory $file = File::Spec->catfile( File::Spec->curdir(), 'temp', 'file.txt' ); # on Unix and Win32, './temp/file.txt' # on Mac OS, ':temp:file.txt'File::Spec is available in the standard distribution, as of version 5.004_05.
In general, production code should not have file paths hardcoded; making them user supplied or from a configuration file is better, keeping in mind that file path syntax varies on different machines.
This is especially noticeable in scripts like Makefiles and test suites, which often assume / as a path separator for subdirectories.
Also of use is File::Basename, from the standard distribution, which splits a pathname into pieces (base filename, full path to directory, and file suffix).
Even when on a single platform (if you can call UNIX a single platform), remember not to count on the existence or the contents of system-specific files, like /etc/passwd, /etc/sendmail.conf, or /etc/resolv.conf. For example the /etc/passwd may exist but it may not contain the encrypted passwords because the system is using some form of enhanced security-- or it may not contain all the accounts because the system is using NIS. If code does need to rely on such a file, include a description of the file and its format in the code's documentation, and make it easy for the user to override the default location of the file.
Do not have two files of the same name with different case, like test.pl and <Test.pl>, as many platforms have case-insensitive filenames. Also, try not to have non-word characters (except for .) in the names, and keep them to the 8.3 convention, for maximum portability.
Likewise, if using AutoSplit, try to keep the split functions to 8.3 naming and case-insensitive conventions; or, at the very least, make it so the resulting files have a unique (case-insensitively) first 8 characters.
Don't assume < won't be the first character of a filename. Always use > explicitly to open a file for reading:
open(FILE, "<$existing_file") or die $!;
Some platforms can't delete or rename files that are being held open by the system. Remember to close files when you are done with them. Don't unlink or rename an open file. Don't tie to or open a file that is already tied to or opened; untie or close first.
Don't open the same file more than once at a time for writing, as some operating systems put mandatory locks on such files.
Don't count on a specific environment variable existing in %ENV. Don't count on %ENV entries being case-sensitive, or even case-preserving.
Don't count on signals.
Don't count on filename globbing. Use opendir, readdir, and closedir instead.
Don't count on per-program environment variables, or per-program current directories.
Commands that launch external processes are generally supported on most platforms (though many of them do not support any type of forking), but the problem with using them arises from what you invoke with them. External tools are often named differently on different platforms, often not available in the same location, often accept different arguments, often behave differently, and often represent their results in a platform-dependent way. Thus you should seldom depend on them to produce consistent results.
One especially common bit of Perl code is opening a pipe to sendmail:
open(MAIL, '|/usr/lib/sendmail -t') or die $!;This is fine for systems programming when sendmail is known to be available. But it is not fine for many non-Unix systems, and even some Unix systems that may not have sendmail installed. If a portable solution is needed, see the Mail::Send and Mail::Mailer modules in the MailTools distribution. Mail::Mailer provides several mailing methods, including mail, sendmail, and direct SMTP (via Net::SMTP) if a mail transfer agent is not available.
The rule of thumb for portable code is: Do it all in portable Perl, or use a module (that may internally implement it with platform-specific code, but expose a common interface).
The UNIX System V IPC (msg*(), sem*(), shm*()) is not available even in all UNIX platforms.
There is a different kind of portability issue with writing XS code: availability of a C compiler on the end-user's system. C brings with it its own portability issues, and writing XS code will expose you to some of those. Writing purely in perl is a comparatively easier way to achieve portability.
There is no one DBM module that is available on all platforms. SDBM_File and the others are generally available on all Unix and DOSish ports, but not in MacPerl, where only NBDM_File and DB_File are available.
The good news is that at least some DBM module should be available, and AnyDBM_File will use whichever module it can find. Of course, then the code needs to be fairly strict, dropping to the lowest common denominator (e.g., not exceeding 1K for each record).
Don't assume that the epoch starts at 00:00:00, January 1, 1970, because that is OS-specific. Better to store a date in an unambiguous representation. The ISO 8601 standard defines YYYY-MM-DD as the date format. A text representation (like 1 Jan 1970) can be easily converted into an OS-specific value using a module like Date::Parse. An array of values, such as those returned by localtime, can be converted to an OS-specific representation using Time::Local.
# NOTE: this is no longer "bad" in perl5.005 for (0..10000000) {} # bad for (my $x = 0; $x <= 10000000; ++$x) {} # good
@lines = <VERY_LARGE_FILE>; # bad
while (<FILE>) {$file .= $_} # sometimes bad $file = join('', <FILE>); # betterThe last two may appear unintuitive to most people. The first of those two constructs repeatedly grows a string, while the second allocates a large chunk of memory in one go. On some systems, the latter is more efficient that the former.
The purpose of the testing is twofold: one, to help developers fix any
problems in their code that crop up because of lack of testing on other
platforms; two, to provide users with information about whether or not
a given module works on a given platform.
uname $^O $Config{'archname'} ------------------------------------------- AIX aix aix FreeBSD freebsd freebsd-i386 Linux linux i386-linux HP-UX hpux PA-RISC1.1 IRIX irix irix OSF1 dec_osf alpha-dec_osf SunOS solaris sun4-solaris SunOS solaris i86pc-solaris SunOS4 sunos sun4-sunosNote that because the $Config{'archname'} may depend on the hardware architecture it may vary quite a lot, much more than the $^O.
$filespec0 = "c:/foo/bar/file.txt"; $filespec1 = "c:\\foo\\bar\\file.txt"; $filespec2 = 'c:\foo\bar\file.txt'; $filespec3 = 'c:\\foo\\bar\\file.txt';System calls accept either / or \ as the path separator. However, many command-line utilities of DOS vintage treat / as the option prefix, so they may get confused by filenames containing /. Aside from calling any external programs, / will work just fine, and probably better, as it is more consistent with popular usage, and avoids the problem of remembering what to backwhack and what not to.
The DOS FAT filesystem can only accommodate ``8.3'' style filenames. Under the ``case insensitive, but case preserving'' HPFS (OS/2) and NTFS (NT) filesystems you may have to be careful about case returned with functions like readdir or used with functions like open or opendir.
DOS also treats several filenames as special, such as AUX, PRN, NUL, CON, COM1, LPT1, LPT2 etc. Unfortunately these filenames won't even work if you include an explicit directory prefix, in some cases. It is best to avoid such filenames, if you want your code to be portable to DOS and its derivatives.
Users of these operating systems may also wish to make use of scripts such as pl2bat.bat or pl2cmd as appropriate to put wrappers around your scripts.
Newline (\n) is translated as \015\012 by STDIO when reading from and writing to files. binmode(FILEHANDLE) will keep \n translated as \012 for that filehandle. Since it is a noop on other systems, binmode should be used for cross-platform code that deals with binary data.
The $^O variable and the $Config{'archname'} values for various DOSish perls are as follows:
OS $^O $Config{'archname'} -------------------------------------------- MS-DOS dos PC-DOS dos OS/2 os2 Windows 95 MSWin32 MSWin32-x86 Windows NT MSWin32 MSWin32-x86 Windows NT MSWin32 MSWin32-alpha Windows NT MSWin32 MSWin32-ppcAlso see:
Directories are specified as:
volume:folder:file for absolute pathnames volume:folder: for absolute pathnames :folder:file for relative pathnames :folder: for relative pathnames :file for relative pathnames file for relative pathnamesFiles in a directory are stored in alphabetical order. Filenames are limited to 31 characters, and may include any character except :, which is reserved as a path separator.
Instead of flock, see FSpSetFLock and FSpRstFLock in the Mac::Files module.
In the MacPerl application, you can't run a program from the command line; programs that expect @ARGV to be populated can be edited with something like the following, which brings up a dialog box asking for the command line arguments.
if (!@ARGV) { @ARGV = split /\s+/, MacPerl::Ask('Arguments?'); }A MacPerl script saved as a droplet will populate @ARGV with the full pathnames of the files dropped onto the script.
Mac users can use programs on a kind of command line under MPW (Macintosh Programmer's Workshop, a free development environment from Apple). MacPerl was first introduced as an MPW tool, and MPW can be used like a shell:
perl myscript.plx some argumentsToolServer is another app from Apple that provides access to MPW tools from MPW and the MacPerl app, which allows MacPerl programs to use system, backticks, and piped open.
``Mac OS'' is the proper name for the operating system, but the value in $^O is ``MacOS''. To determine architecture, version, or whether the application or MPW tool version is running, check:
$is_app = $MacPerl::Version =~ /App/; $is_tool = $MacPerl::Version =~ /MPW/; ($version) = $MacPerl::Version =~ /^(\S+)/; $is_ppc = $MacPerl::Architecture eq 'MacPPC'; $is_68k = $MacPerl::Architecture eq 'Mac68K';Mac OS X, to be based on NeXT's OpenStep OS, will be able to run MacPerl natively (in the Blue Box, and even in the Yellow Box, once some changes to the toolbox calls are made), but Unix perl will also run natively.
Also see:
$ perl -ne "print if /perl_setup/i" SYS$LOGIN:LOGIN.COM $ perl -ne "print if /perl_setup/i" /sys$login/login.combut not a mixture of both as in:
$ perl -ne "print if /perl_setup/i" sys$login:/login.com Can't open sys$login:/login.com: file specification syntax errorInteracting with Perl from the Digital Command Language (DCL) shell often requires a different set of quotation marks than Unix shells do. For example:
$ perl -e "print ""Hello, world.\n""" Hello, world.There are a number of ways to wrap your perl scripts in DCL .COM files if you are so inclined. For example:
$ write sys$output "Hello from DCL!" $ if p1 .eqs. "" $ then perl -x 'f$environment("PROCEDURE") $ else perl -x - 'p1 'p2 'p3 'p4 'p5 'p6 'p7 'p8 $ deck/dollars="__END__" #!/usr/bin/perl
print "Hello from Perl!\n";
__END__ $ endifDo take care with $ ASSIGN/nolog/user SYS$COMMAND: SYS$INPUT if your perl-in-DCL script expects to do things like $read = <STDIN>;.
Filenames are in the format ``name.extension;version''. The maximum length for filenames is 39 characters, and the maximum length for extensions is also 39 characters. Version is a number from 1 to 32767. Valid characters are /[A-Z0-9$_-]/.
VMS' RMS filesystem is case insensitive and does not preserve case. readdir returns lowercased filenames, but specifying a file for opening remains case insensitive. Files without extensions have a trailing period on them, so doing a readdir with a file named A.;5 will return a. (though that file could be opened with open(FH, 'A')).
RMS had an eight level limit on directory depths from any rooted logical (allowing 16 levels overall) prior to VMS 7.2. Hence PERL_ROOT:[LIB.2.3.4.5.6.7.8] is a valid directory specification but PERL_ROOT:[LIB.2.3.4.5.6.7.8.9] is not. Makefile.PL authors might have to take this into account, but at least they can refer to the former as /PERL_ROOT/lib/2/3/4/5/6/7/8/.
The VMS::Filespec module, which gets installed as part of the build process on VMS, is a pure Perl module that can easily be installed on non-VMS platforms and can be helpful for conversions to and from RMS native formats.
What \n represents depends on the type of file that is open. It could be \015, \012, \015\012, or nothing. Reading from a file translates newlines to \012, unless binmode was executed on that handle, just like DOSish perls.
TCP/IP stacks are optional on VMS, so socket routines might not be implemented. UDP sockets may not be supported.
The value of $^O on OpenVMS is ``VMS''. To determine the architecture that you are running on without resorting to loading all of %Config you can examine the content of the @INC array like so:
if (grep(/VMS_AXP/, @INC)) { print "I'm on Alpha!\n"; } elsif (grep(/VMS_VAX/, @INC)) { print "I'm on VAX!\n"; } else { print "I'm not so sure about where $^O is...\n"; }Also see:
As of R2.5 of USS for OS/390 that Unix sub-system did not support the #! shebang trick for script invocation. Hence, on OS/390 perl scripts can executed with a header similar to the following simple script:
: # use perl eval 'exec /usr/local/bin/perl -S $0 ${1+"$@"}' if 0; #!/usr/local/bin/perl # just a comment really
print "Hello from perl!\n";On these platforms, bear in mind that the EBCDIC character set may have an effect on what happens with some perl functions (such as chr, pack, print, printf, ord, sort, sprintf, unpack), as well as bit-fiddling with ASCII constants using operators like ^, & and |, not to mention dealing with socket interfaces to ASCII computers (see the section on NEWLINES).
Fortunately, most web servers for the mainframe will correctly translate the \n in the following statement to its ASCII equivalent (note that \r is the same under both Unix and OS/390):
print "\r\n\r\n";The value of $^O on OS/390 is ``os390''.
Some simple tricks for determining if you are running on an EBCDIC platform could include any of the following (perhaps all):
if ("\t" eq "\05") { print "EBCDIC may be spoken here!\n"; }
if (ord('A') == 193) { print "EBCDIC may be spoken here!\n"; }
if (chr(169) eq 'z') { print "EBCDIC may be spoken here!\n"; }Note that one thing you may not want to rely on is the EBCDIC encoding of punctuation characters since these may differ from code page to code page (and once your module or script is rumoured to work with EBCDIC, folks will want it to work with all EBCDIC character sets).
Also see:
Native filenames are of the form
Filesystem#Special_Field::DiscName.$.Directory.Directory.Filewhere
Special_Field is not usually present, but may contain . and $ . Filesystem =~ m|[A-Za-z0-9_]| DsicName =~ m|[A-Za-z0-9_/]| $ represents the root directory . is the path separator @ is the current directory (per filesystem but machine global) ^ is the parent directory Directory and File =~ m|[^\0- "\.\$\%\&:\@\\^\|\177]+|The default filename translation is roughly tr|/.|./|;
Note that "ADFS::HardDisc.$.File" ne 'ADFS::HardDisc.$.File' and that the second stage of $ interpolation in regular expressions will fall foul of the $. if scripts are not careful.
Logical paths specified by system variables containing comma-separated search lists are also allowed, hence System:Modules is a valid filename, and the filesystem will prefix Modules with each section of System$Path until a name is made that points to an object on disc. Writing to a new file System:Modules would only be allowed if System$Path contains a single item list. The filesystem will also expand system variables in filenames if enclosed in angle brackets, so <System$Dir>.Modules would look for the file $ENV{'System$Dir'} . 'Modules'. The obvious implication of this is that B<fully qualified filenames can start with <> and should be protected when open is used for input.
Because . was in use as a directory separator and filenames could not be assumed to be unique after 10 characters, Acorn implemented the C compiler to strip the trailing .c .h .s and .o suffix from filenames specified in source code and store the respective files in subdirectories named after the suffix. Hence files are translated:
foo.h h.foo C:foo.h C:h.foo (logical path variable) sys/os.h sys.h.os (C compiler groks Unix-speak) 10charname.c c.10charname 10charname.o o.10charname 11charname_.c c.11charname (assuming filesystem truncates at 10)The Unix emulation library's translation of filenames to native assumes that this sort of translation is required, and allows a user defined list of known suffixes which it will transpose in this fashion. This may appear transparent, but consider that with these rules foo/bar/baz.h and foo/bar/h/baz both map to foo.bar.h.baz, and that readdir and glob cannot and do not attempt to emulate the reverse mapping. Other .s in filenames are translated to /.
As implied above the environment accessed through %ENV is global, and the convention is that program specific environment variables are of the form Program$Name. Each filing system maintains a current directory, and the current filing system's current directory is the global current directory. Consequently, sociable scripts don't change the current directory but rely on full pathnames, and scripts (and Makefiles) cannot assume that they can spawn a child process which can change the current directory without affecting its parent (and everyone else for that matter).
As native operating system filehandles are global and currently are allocated down from 255, with 0 being a reserved value the Unix emulation library emulates Unix filehandles. Consequently, you can't rely on passing STDIN, STDOUT, or STDERR to your children.
The desire of users to express filenames of the form <Foo$Dir>.Bar on the command line unquoted causes problems, too: `` command output capture has to perform a guessing game. It assumes that a string <[^<>]+\$[^<>]> is a reference to an environment variable, whereas anything else involving < or > is redirection, and generally manages to be 99% right. Of course, the problem remains that scripts cannot rely on any Unix tools being available, or that any tools found have Unix-like command line arguments.
Extensions and XS are, in theory, buildable by anyone using free tools. In practice, many don't, as users of the Acorn platform are used to binary distribution. MakeMaker does run, but no available make currently copes with MakeMaker's makefiles; even if/when this is fixed, the lack of a Unix-like shell can cause problems with makefile rules, especially lines of the form cd sdbm && make all, and anything using quoting.
``RISC OS'' is the proper name for the operating system, but the value in $^O is ``riscos'' (because we don't like shouting).
Also see:
See also:
The list may very well be incomplete, or wrong in some places. When in doubt, consult the platform-specific README files in the Perl source distribution, and other documentation resources for a given port.
Be aware, moreover, that even among Unix-ish systems there are variations.
For many functions, you can also query %Config, exported by default from Config.pm. For example, to check if the platform has the lstat call, check $Config{'d_lstat'}. See the Config.pm manpage for a full description of available variables.
-r, -w, -x, and -o tell whether or not file is accessible, which may not reflect UIC-based file protections. (VMS)
-s returns the size of the data fork, not the total size of data fork plus resource fork. (Mac OS).
-s by name on an open file will return the space reserved on disk, rather than the current extent. -s on an open filehandle returns the current size. (RISC OS)
-R, -W, -X, -O are indistinguishable from -r, -w, -x, -o. (Mac OS, Win32, VMS, RISC OS)
-b, -c, -k, -g, -p, -u, -A are not implemented. (Mac OS)
-g, -k, -l, -p, -u, -A are not particularly meaningful. (Win32, VMS, RISC OS)
-d is true if passed a device spec without an explicit directory. (VMS)
-T and -B are implemented, but might misclassify Mac text files with foreign characters; this is the case will all platforms, but may affect Mac OS often. (Mac OS)
-x (or -X) determine if a file ends in one of the executable suffixes. -S is meaningless. (Win32)
-x (or -X) determine if a file has an executable file type.
(RISC OS)
Reopens file and restores pointer; if function fails, underlying filehandle may be closed, or pointer may be in a different position. (VMS)
The value returned by tell may be affected after the call, and
the filehandle may be flushed. (Win32)
Only good for changing ``owner'' read-write access, ``group'', and ``other'' bits are meaningless. (Win32)
Only good for changing ``owner'' and ``other'' read-write access. (RISC OS)
Does nothing, but won't fail. (Win32)
Not implemented. (Win32)
Invokes VMS debugger. (VMS)
Available only on Windows NT (not on Windows 95). (Win32)
Not useful. (RISC OS)
Not useful. (RISC OS)
Features depend on external perlglob.exe or perlglob.bat. May be overridden with something like File::DosGlob, which is recommended. (Win32)
Globbing built-in, but only * and ? metacharacters are supported.
Globbing relies on operating system calls, which may return filenames
in any order. As most filesystems are case-insensitive, even ``sorted''
filenames will not be in case-sensitive order. (RISC OS)
Available only for socket handles, and it does what the ioctlsocket() call in the Winsock API does. (Win32)
Available only for socket handles. (RISC OS)
Available only for process handles returned by the system(1, ...)
method of spawning a process. (Win32)
Return values may be bogus. (Win32)
open to |- and -| are unsupported. (Mac OS, Win32, RISC OS)
Only reliable on sockets. (RISC OS)
device and inode are not meaningful. (Win32)
device and inode are not necessarily reliable. (VMS)
mtime, atime and ctime all return the last modification time. Device and
inode are not necessarily reliable. (RISC OS)
As an optimization, may not call the command shell specified in $ENV{PERL5SHELL}. system(1, @args) spawns an external process and immediately returns its process designator, without waiting for it to terminate. Return value may be used subsequently in wait or waitpid. (Win32)
There is no shell to process metacharacters, and the native standard is
to pass a command line terminated by ``\n" ``\r'' or ``\0'' to the spawned
program. Redirection such as > foo is performed (if at all) by
the run time library of the spawned program. system list will call
the Unix emulation library's exec emulation, which attempts to provide
emulation of the stdin, stdout, stderr in force in the parent, providing
the child program uses a compatible version of the emulation library.
scalar will call the native command line direct and no such emulation
of a child Unix program will exists. Mileage will vary. (RISC OS)
``cumulative'' times will be bogus. On anything other than Windows NT, ``system'' time will be bogus, and ``user'' time is actually the time returned by the clock() function in the C runtime library. (Win32)
Not useful. (RISC OS)
May not behave as expected. Behavior depends on the C runtime
library's implementation of utime(), and the filesystem being
used. The FAT filesystem typically does not support an ``access
time'' field, and it may limit timestamps to a granularity of
two seconds. (Win32)
Can only be applied to process handles returned for processes spawned using system(1, ...). (Win32)
Not useful. (RISC OS)
This document is maintained by Chris Nandor.