More perldelta rototilling.

[perl5.git] / pod / perlfaq8.pod

=head1 NAME

perlfaq8 - System Interaction

=head1 DESCRIPTION

This section of the Perl FAQ covers questions involving operating
system interaction.  Topics include interprocess communication (IPC),
control over the user-interface (keyboard, screen and pointing
devices), and most anything else not related to data manipulation.

Read the FAQs and documentation specific to the port of perl to your
operating system (eg, L<perlvms>, L<perlplan9>, ...).  These should
contain more detailed information on the vagaries of your perl.

=head2 How do I find out which operating system I'm running under?

The $^O variable ($OSNAME if you use English) contains an indication of
the name of the operating system (not its release number) that your perl
binary was built for.

=head2 How come exec() doesn't return?
X<exec> X<system> X<fork> X<open> X<pipe>

(contributed by brian d foy)

The C<exec> function's job is to turn your process into another
command and never to return. If that's not what you want to do, don't
use C<exec>. :)

If you want to run an external command and still keep your Perl process
going, look at a piped C<open>, C<fork>, or C<system>.

=head2 How do I do fancy stuff with the keyboard/screen/mouse?

How you access/control keyboards, screens, and pointing devices
("mice") is system-dependent.  Try the following modules:

=over 4

=item Keyboard

        Term::Cap               Standard perl distribution
        Term::ReadKey           CPAN
        Term::ReadLine::Gnu     CPAN
        Term::ReadLine::Perl    CPAN
        Term::Screen            CPAN

=item Screen

        Term::Cap               Standard perl distribution
        Curses                  CPAN
        Term::ANSIColor         CPAN

=item Mouse

        Tk                      CPAN

=back

Some of these specific cases are shown as examples in other answers
in this section of the perlfaq.

=head2 How do I print something out in color?

In general, you don't, because you don't know whether
the recipient has a color-aware display device.  If you
know that they have an ANSI terminal that understands
color, you can use the Term::ANSIColor module from CPAN:

        use Term::ANSIColor;
        print color("red"), "Stop!\n", color("reset");
        print color("green"), "Go!\n", color("reset");

Or like this:

        use Term::ANSIColor qw(:constants);
        print RED, "Stop!\n", RESET;
        print GREEN, "Go!\n", RESET;

=head2 How do I read just one key without waiting for a return key?

Controlling input buffering is a remarkably system-dependent matter.
On many systems, you can just use the B<stty> command as shown in
L<perlfunc/getc>, but as you see, that's already getting you into
portability snags.

        open(TTY, "+</dev/tty") or die "no tty: $!";
        system "stty  cbreak </dev/tty >/dev/tty 2>&1";
        $key = getc(TTY);               # perhaps this works
        # OR ELSE
        sysread(TTY, $key, 1);  # probably this does
        system "stty -cbreak </dev/tty >/dev/tty 2>&1";

The Term::ReadKey module from CPAN offers an easy-to-use interface that
should be more efficient than shelling out to B<stty> for each key.
It even includes limited support for Windows.

        use Term::ReadKey;
        ReadMode('cbreak');
        $key = ReadKey(0);
        ReadMode('normal');

However, using the code requires that you have a working C compiler
and can use it to build and install a CPAN module.  Here's a solution
using the standard POSIX module, which is already on your systems
(assuming your system supports POSIX).

        use HotKey;
        $key = readkey();

And here's the HotKey module, which hides the somewhat mystifying calls
to manipulate the POSIX termios structures.

        # HotKey.pm
        package HotKey;

        @ISA = qw(Exporter);
        @EXPORT = qw(cbreak cooked readkey);

        use strict;
        use POSIX qw(:termios_h);
        my ($term, $oterm, $echo, $noecho, $fd_stdin);

        $fd_stdin = fileno(STDIN);
        $term     = POSIX::Termios->new();
        $term->getattr($fd_stdin);
        $oterm     = $term->getlflag();

        $echo     = ECHO | ECHOK | ICANON;
        $noecho   = $oterm & ~$echo;

        sub cbreak {
                $term->setlflag($noecho);  # ok, so i don't want echo either
                $term->setcc(VTIME, 1);
                $term->setattr($fd_stdin, TCSANOW);
        }

        sub cooked {
                $term->setlflag($oterm);
                $term->setcc(VTIME, 0);
                $term->setattr($fd_stdin, TCSANOW);
        }

        sub readkey {
                my $key = '';
                cbreak();
                sysread(STDIN, $key, 1);
                cooked();
                return $key;
        }

        END { cooked() }

        1;

=head2 How do I check whether input is ready on the keyboard?

The easiest way to do this is to read a key in nonblocking mode with the
Term::ReadKey module from CPAN, passing it an argument of -1 to indicate
not to block:

        use Term::ReadKey;

        ReadMode('cbreak');

        if (defined ($char = ReadKey(-1)) ) {
                # input was waiting and it was $char
        } else {
                # no input was waiting
        }

        ReadMode('normal');                  # restore normal tty settings

=head2 How do I clear the screen?

(contributed by brian d foy)

To clear the screen, you just have to print the special sequence
that tells the terminal to clear the screen. Once you have that
sequence, output it when you want to clear the screen.

You can use the C<Term::ANSIScreen> module to get the special
sequence. Import the C<cls> function (or the C<:screen> tag):

        use Term::ANSIScreen qw(cls);
        my $clear_screen = cls();

        print $clear_screen;

The C<Term::Cap> module can also get the special sequence if you want
to deal with the low-level details of terminal control. The C<Tputs>
method returns the string for the given capability:

        use Term::Cap;

        $terminal = Term::Cap->Tgetent( { OSPEED => 9600 } );
        $clear_string = $terminal->Tputs('cl');

        print $clear_screen;

On Windows, you can use the C<Win32::Console> module. After creating
an object for the output filehandle you want to affect, call the
C<Cls> method:

        Win32::Console;

        $OUT = Win32::Console->new(STD_OUTPUT_HANDLE);
        my $clear_string = $OUT->Cls;

        print $clear_screen;

If you have a command-line program that does the job, you can call
it in backticks to capture whatever it outputs so you can use it
later:

        $clear_string = `clear`;

        print $clear_string;

=head2 How do I get the screen size?

If you have Term::ReadKey module installed from CPAN,
you can use it to fetch the width and height in characters
and in pixels:

        use Term::ReadKey;
        ($wchar, $hchar, $wpixels, $hpixels) = GetTerminalSize();

This is more portable than the raw C<ioctl>, but not as
illustrative:

        require 'sys/ioctl.ph';
        die "no TIOCGWINSZ " unless defined &TIOCGWINSZ;
        open(TTY, "+</dev/tty")                     or die "No tty: $!";
        unless (ioctl(TTY, &TIOCGWINSZ, $winsize='')) {
                die sprintf "$0: ioctl TIOCGWINSZ (%08x: $!)\n", &TIOCGWINSZ;
        }
        ($row, $col, $xpixel, $ypixel) = unpack('S4', $winsize);
        print "(row,col) = ($row,$col)";
        print "  (xpixel,ypixel) = ($xpixel,$ypixel)" if $xpixel || $ypixel;
        print "\n";

=head2 How do I ask the user for a password?

(This question has nothing to do with the web.  See a different
FAQ for that.)

There's an example of this in L<perlfunc/crypt>).  First, you put the
terminal into "no echo" mode, then just read the password normally.
You may do this with an old-style ioctl() function, POSIX terminal
control (see L<POSIX> or its documentation the Camel Book), or a call
to the B<stty> program, with varying degrees of portability.

You can also do this for most systems using the Term::ReadKey module
from CPAN, which is easier to use and in theory more portable.

        use Term::ReadKey;

        ReadMode('noecho');
        $password = ReadLine(0);

=head2 How do I read and write the serial port?

This depends on which operating system your program is running on.  In
the case of Unix, the serial ports will be accessible through files in
/dev; on other systems, device names will doubtless differ.
Several problem areas common to all device interaction are the
following:

=over 4

=item lockfiles

Your system may use lockfiles to control multiple access.  Make sure
you follow the correct protocol.  Unpredictable behavior can result
from multiple processes reading from one device.

=item open mode

If you expect to use both read and write operations on the device,
you'll have to open it for update (see L<perlfunc/"open"> for
details).  You may wish to open it without running the risk of
blocking by using sysopen() and C<O_RDWR|O_NDELAY|O_NOCTTY> from the
Fcntl module (part of the standard perl distribution).  See
L<perlfunc/"sysopen"> for more on this approach.

=item end of line

Some devices will be expecting a "\r" at the end of each line rather
than a "\n".  In some ports of perl, "\r" and "\n" are different from
their usual (Unix) ASCII values of "\012" and "\015".  You may have to
give the numeric values you want directly, using octal ("\015"), hex
("0x0D"), or as a control-character specification ("\cM").

        print DEV "atv1\012";   # wrong, for some devices
        print DEV "atv1\015";   # right, for some devices

Even though with normal text files a "\n" will do the trick, there is
still no unified scheme for terminating a line that is portable
between Unix, DOS/Win, and Macintosh, except to terminate I<ALL> line
ends with "\015\012", and strip what you don't need from the output.
This applies especially to socket I/O and autoflushing, discussed
next.

=item flushing output

If you expect characters to get to your device when you print() them,
you'll want to autoflush that filehandle.  You can use select()
and the C<$|> variable to control autoflushing (see L<perlvar/$E<verbar>>
and L<perlfunc/select>, or L<perlfaq5>, "How do I flush/unbuffer an
output filehandle?  Why must I do this?"):

        $oldh = select(DEV);
        $| = 1;
        select($oldh);

You'll also see code that does this without a temporary variable, as in

        select((select(DEV), $| = 1)[0]);

Or if you don't mind pulling in a few thousand lines
of code just because you're afraid of a little $| variable:

        use IO::Handle;
        DEV->autoflush(1);

As mentioned in the previous item, this still doesn't work when using
socket I/O between Unix and Macintosh.  You'll need to hard code your
line terminators, in that case.

=item non-blocking input

If you are doing a blocking read() or sysread(), you'll have to
arrange for an alarm handler to provide a timeout (see
L<perlfunc/alarm>).  If you have a non-blocking open, you'll likely
have a non-blocking read, which means you may have to use a 4-arg
select() to determine whether I/O is ready on that device (see
L<perlfunc/"select">.

=back

While trying to read from his caller-id box, the notorious Jamie Zawinski
C<< <jwz@netscape.com> >>, after much gnashing of teeth and fighting with sysread,
sysopen, POSIX's tcgetattr business, and various other functions that
go bump in the night, finally came up with this:

        sub open_modem {
                use IPC::Open2;
                my $stty = `/bin/stty -g`;
                open2( \*MODEM_IN, \*MODEM_OUT, "cu -l$modem_device -s2400 2>&1");
                # starting cu hoses /dev/tty's stty settings, even when it has
                # been opened on a pipe...
                system("/bin/stty $stty");
                $_ = <MODEM_IN>;
                chomp;
                if ( !m/^Connected/ ) {
                        print STDERR "$0: cu printed `$_' instead of `Connected'\n";
                }
        }

=head2 How do I decode encrypted password files?

You spend lots and lots of money on dedicated hardware, but this is
bound to get you talked about.

Seriously, you can't if they are Unix password files--the Unix
password system employs one-way encryption.  It's more like hashing
than encryption.  The best you can do is check whether something else
hashes to the same string.  You can't turn a hash back into the
original string. Programs like Crack can forcibly (and intelligently)
try to guess passwords, but don't (can't) guarantee quick success.

If you're worried about users selecting bad passwords, you should
proactively check when they try to change their password (by modifying
passwd(1), for example).

=head2 How do I start a process in the background?

(contributed by brian d foy)

There's not a single way to run code in the background so you don't
have to wait for it to finish before your program moves on to other
tasks. Process management depends on your particular operating system,
and many of the techniques are in L<perlipc>.

Several CPAN modules may be able to help, including IPC::Open2 or
IPC::Open3, IPC::Run, Parallel::Jobs, Parallel::ForkManager, POE,
Proc::Background, and Win32::Process. There are many other modules you
might use, so check those namespaces for other options too.

If you are on a unix-like system, you might be able to get away with a
system call where you put an C<&> on the end of the command:

        system("cmd &")

You can also try using C<fork>, as described in L<perlfunc> (although
this is the same thing that many of the modules will do for you).

=over 4

=item STDIN, STDOUT, and STDERR are shared

Both the main process and the backgrounded one (the "child" process)
share the same STDIN, STDOUT and STDERR filehandles.  If both try to
access them at once, strange things can happen.  You may want to close
or reopen these for the child.  You can get around this with
C<open>ing a pipe (see L<perlfunc/"open">) but on some systems this
means that the child process cannot outlive the parent.

=item Signals

You'll have to catch the SIGCHLD signal, and possibly SIGPIPE too.
SIGCHLD is sent when the backgrounded process finishes.  SIGPIPE is
sent when you write to a filehandle whose child process has closed (an
untrapped SIGPIPE can cause your program to silently die).  This is
not an issue with C<system("cmd&")>.

=item Zombies

You have to be prepared to "reap" the child process when it finishes.

        $SIG{CHLD} = sub { wait };

        $SIG{CHLD} = 'IGNORE';

You can also use a double fork. You immediately wait() for your
first child, and the init daemon will wait() for your grandchild once
it exits.

        unless ($pid = fork) {
            unless (fork) {
                exec "what you really wanna do";
                die "exec failed!";
            }
            exit 0;
        }
        waitpid($pid, 0);

See L<perlipc/"Signals"> for other examples of code to do this.
Zombies are not an issue with C<system("prog &")>.

=back

=head2 How do I trap control characters/signals?

You don't actually "trap" a control character.  Instead, that character
generates a signal which is sent to your terminal's currently
foregrounded process group, which you then trap in your process.
Signals are documented in L<perlipc/"Signals"> and the
section on "Signals" in the Camel.

You can set the values of the %SIG hash to be the functions you want
to handle the signal.  After perl catches the signal, it looks in %SIG
for a key with the same name as the signal, then calls the subroutine
value for that key.

        # as an anonymous subroutine

        $SIG{INT} = sub { syswrite(STDERR, "ouch\n", 5 ) };

        # or a reference to a function

        $SIG{INT} = \&ouch;

        # or the name of the function as a string

        $SIG{INT} = "ouch";

Perl versions before 5.8 had in its C source code signal handlers which
would catch the signal and possibly run a Perl function that you had set
in %SIG.  This violated the rules of signal handling at that level
causing perl to dump core. Since version 5.8.0, perl looks at %SIG
*after* the signal has been caught, rather than while it is being caught.
Previous versions of this answer were incorrect.

=head2 How do I modify the shadow password file on a Unix system?

If perl was installed correctly and your shadow library was written
properly, the getpw*() functions described in L<perlfunc> should in
theory provide (read-only) access to entries in the shadow password
file.  To change the file, make a new shadow password file (the format
varies from system to system--see L<passwd> for specifics) and use
pwd_mkdb(8) to install it (see L<pwd_mkdb> for more details).

=head2 How do I set the time and date?

Assuming you're running under sufficient permissions, you should be
able to set the system-wide date and time by running the date(1)
program.  (There is no way to set the time and date on a per-process
basis.)  This mechanism will work for Unix, MS-DOS, Windows, and NT;
the VMS equivalent is C<set time>.

However, if all you want to do is change your time zone, you can
probably get away with setting an environment variable:

        $ENV{TZ} = "MST7MDT";              # unixish
        $ENV{'SYS$TIMEZONE_DIFFERENTIAL'}="-5" # vms
        system "trn comp.lang.perl.misc";

=head2 How can I sleep() or alarm() for under a second?
X<Time::HiRes> X<BSD::Itimer> X<sleep> X<select>

If you want finer granularity than the 1 second that the C<sleep()>
function provides, the easiest way is to use the C<select()> function as
documented in L<perlfunc/"select">.  Try the C<Time::HiRes> and
the C<BSD::Itimer> modules (available from CPAN, and starting from
Perl 5.8 C<Time::HiRes> is part of the standard distribution).

=head2 How can I measure time under a second?
X<Time::HiRes> X<BSD::Itimer> X<sleep> X<select>

(contributed by brian d foy)

The C<Time::HiRes> module (part of the standard distribution as of
Perl 5.8) measures time with the C<gettimeofday()> system call, which
returns the time in microseconds since the epoch. If you can't install
C<Time::HiRes> for older Perls and you are on a Unixish system, you
may be able to call C<gettimeofday(2)> directly. See
L<perlfunc/syscall>.

=head2 How can I do an atexit() or setjmp()/longjmp()? (Exception handling)

Release 5 of Perl added the END block, which can be used to simulate
atexit().  Each package's END block is called when the program or
thread ends (see L<perlmod> manpage for more details).

For example, you can use this to make sure your filter program
managed to finish its output without filling up the disk:

        END {
                close(STDOUT) || die "stdout close failed: $!";
        }

The END block isn't called when untrapped signals kill the program,
though, so if you use END blocks you should also use

        use sigtrap qw(die normal-signals);

Perl's exception-handling mechanism is its eval() operator.  You can
use eval() as setjmp and die() as longjmp.  For details of this, see
the section on signals, especially the time-out handler for a blocking
flock() in L<perlipc/"Signals"> or the section on "Signals" in
the Camel Book.

If exception handling is all you're interested in, try the
exceptions.pl library (part of the standard perl distribution).

If you want the atexit() syntax (and an rmexit() as well), try the
AtExit module available from CPAN.

=head2 Why doesn't my sockets program work under System V (Solaris)?  What does the error message "Protocol not supported" mean?

Some Sys-V based systems, notably Solaris 2.X, redefined some of the
standard socket constants.  Since these were constant across all
architectures, they were often hardwired into perl code.  The proper
way to deal with this is to "use Socket" to get the correct values.

Note that even though SunOS and Solaris are binary compatible, these
values are different.  Go figure.

=head2 How can I call my system's unique C functions from Perl?

In most cases, you write an external module to do it--see the answer
to "Where can I learn about linking C with Perl? [h2xs, xsubpp]".
However, if the function is a system call, and your system supports
syscall(), you can use the syscall function (documented in
L<perlfunc>).

Remember to check the modules that came with your distribution, and
CPAN as well--someone may already have written a module to do it. On
Windows, try Win32::API.  On Macs, try Mac::Carbon.  If no module
has an interface to the C function, you can inline a bit of C in your
Perl source with Inline::C.

=head2 Where do I get the include files to do ioctl() or syscall()?

Historically, these would be generated by the h2ph tool, part of the
standard perl distribution.  This program converts cpp(1) directives
in C header files to files containing subroutine definitions, like
&SYS_getitimer, which you can use as arguments to your functions.
It doesn't work perfectly, but it usually gets most of the job done.
Simple files like F<errno.h>, F<syscall.h>, and F<socket.h> were fine,
but the hard ones like F<ioctl.h> nearly always need to hand-edited.
Here's how to install the *.ph files:

        1.  become super-user
        2.  cd /usr/include
        3.  h2ph *.h */*.h

If your system supports dynamic loading, for reasons of portability and
sanity you probably ought to use h2xs (also part of the standard perl
distribution).  This tool converts C header files to Perl extensions.
See L<perlxstut> for how to get started with h2xs.

If your system doesn't support dynamic loading, you still probably
ought to use h2xs.  See L<perlxstut> and L<ExtUtils::MakeMaker> for
more information (in brief, just use B<make perl> instead of a plain
B<make> to rebuild perl with a new static extension).

=head2 Why do setuid perl scripts complain about kernel problems?

Some operating systems have bugs in the kernel that make setuid
scripts inherently insecure.  Perl gives you a number of options
(described in L<perlsec>) to work around such systems.

=head2 How can I open a pipe both to and from a command?

The IPC::Open2 module (part of the standard perl distribution) is an
easy-to-use approach that internally uses pipe(), fork(), and exec() to do
the job.  Make sure you read the deadlock warnings in its documentation,
though (see L<IPC::Open2>).  See
L<perlipc/"Bidirectional Communication with Another Process"> and
L<perlipc/"Bidirectional Communication with Yourself">

You may also use the IPC::Open3 module (part of the standard perl
distribution), but be warned that it has a different order of
arguments from IPC::Open2 (see L<IPC::Open3>).

=head2 Why can't I get the output of a command with system()?

You're confusing the purpose of system() and backticks (``).  system()
runs a command and returns exit status information (as a 16 bit value:
the low 7 bits are the signal the process died from, if any, and
the high 8 bits are the actual exit value).  Backticks (``) run a
command and return what it sent to STDOUT.

        $exit_status   = system("mail-users");
        $output_string = `ls`;

=head2 How can I capture STDERR from an external command?

There are three basic ways of running external commands:

        system $cmd;            # using system()
        $output = `$cmd`;               # using backticks (``)
        open (PIPE, "cmd |");   # using open()

With system(), both STDOUT and STDERR will go the same place as the
script's STDOUT and STDERR, unless the system() command redirects them.
Backticks and open() read B<only> the STDOUT of your command.

You can also use the open3() function from IPC::Open3.  Benjamin
Goldberg provides some sample code:

To capture a program's STDOUT, but discard its STDERR:

        use IPC::Open3;
        use File::Spec;
        use Symbol qw(gensym);
        open(NULL, ">", File::Spec->devnull);
        my $pid = open3(gensym, \*PH, ">&NULL", "cmd");
        while( <PH> ) { }
        waitpid($pid, 0);

To capture a program's STDERR, but discard its STDOUT:

        use IPC::Open3;
        use File::Spec;
        use Symbol qw(gensym);
        open(NULL, ">", File::Spec->devnull);
        my $pid = open3(gensym, ">&NULL", \*PH, "cmd");
        while( <PH> ) { }
        waitpid($pid, 0);

To capture a program's STDERR, and let its STDOUT go to our own STDERR:

        use IPC::Open3;
        use Symbol qw(gensym);
        my $pid = open3(gensym, ">&STDERR", \*PH, "cmd");
        while( <PH> ) { }
        waitpid($pid, 0);

To read both a command's STDOUT and its STDERR separately, you can
redirect them to temp files, let the command run, then read the temp
files:

        use IPC::Open3;
        use Symbol qw(gensym);
        use IO::File;
        local *CATCHOUT = IO::File->new_tmpfile;
        local *CATCHERR = IO::File->new_tmpfile;
        my $pid = open3(gensym, ">&CATCHOUT", ">&CATCHERR", "cmd");
        waitpid($pid, 0);
        seek $_, 0, 0 for \*CATCHOUT, \*CATCHERR;
        while( <CATCHOUT> ) {}
        while( <CATCHERR> ) {}

But there's no real need for *both* to be tempfiles... the following
should work just as well, without deadlocking:

        use IPC::Open3;
        use Symbol qw(gensym);
        use IO::File;
        local *CATCHERR = IO::File->new_tmpfile;
        my $pid = open3(gensym, \*CATCHOUT, ">&CATCHERR", "cmd");
        while( <CATCHOUT> ) {}
        waitpid($pid, 0);
        seek CATCHERR, 0, 0;
        while( <CATCHERR> ) {}

And it'll be faster, too, since we can begin processing the program's
stdout immediately, rather than waiting for the program to finish.

With any of these, you can change file descriptors before the call:

        open(STDOUT, ">logfile");
        system("ls");

or you can use Bourne shell file-descriptor redirection:

        $output = `$cmd 2>some_file`;
        open (PIPE, "cmd 2>some_file |");

You can also use file-descriptor redirection to make STDERR a
duplicate of STDOUT:

        $output = `$cmd 2>&1`;
        open (PIPE, "cmd 2>&1 |");

Note that you I<cannot> simply open STDERR to be a dup of STDOUT
in your Perl program and avoid calling the shell to do the redirection.
This doesn't work:

        open(STDERR, ">&STDOUT");
        $alloutput = `cmd args`;  # stderr still escapes

This fails because the open() makes STDERR go to where STDOUT was
going at the time of the open().  The backticks then make STDOUT go to
a string, but don't change STDERR (which still goes to the old
STDOUT).

Note that you I<must> use Bourne shell (sh(1)) redirection syntax in
backticks, not csh(1)!  Details on why Perl's system() and backtick
and pipe opens all use the Bourne shell are in the
F<versus/csh.whynot> article in the "Far More Than You Ever Wanted To
Know" collection in http://www.cpan.org/misc/olddoc/FMTEYEWTK.tgz .  To
capture a command's STDERR and STDOUT together:

        $output = `cmd 2>&1`;                       # either with backticks
        $pid = open(PH, "cmd 2>&1 |");              # or with an open pipe
        while (<PH>) { }                            #    plus a read

To capture a command's STDOUT but discard its STDERR:

        $output = `cmd 2>/dev/null`;                # either with backticks
        $pid = open(PH, "cmd 2>/dev/null |");       # or with an open pipe
        while (<PH>) { }                            #    plus a read

To capture a command's STDERR but discard its STDOUT:

        $output = `cmd 2>&1 1>/dev/null`;           # either with backticks
        $pid = open(PH, "cmd 2>&1 1>/dev/null |");  # or with an open pipe
        while (<PH>) { }                            #    plus a read

To exchange a command's STDOUT and STDERR in order to capture the STDERR
but leave its STDOUT to come out our old STDERR:

        $output = `cmd 3>&1 1>&2 2>&3 3>&-`;        # either with backticks
        $pid = open(PH, "cmd 3>&1 1>&2 2>&3 3>&-|");# or with an open pipe
        while (<PH>) { }                            #    plus a read

To read both a command's STDOUT and its STDERR separately, it's easiest
to redirect them separately to files, and then read from those files
when the program is done:

        system("program args 1>program.stdout 2>program.stderr");

Ordering is important in all these examples.  That's because the shell
processes file descriptor redirections in strictly left to right order.

        system("prog args 1>tmpfile 2>&1");
        system("prog args 2>&1 1>tmpfile");

The first command sends both standard out and standard error to the
temporary file.  The second command sends only the old standard output
there, and the old standard error shows up on the old standard out.

=head2 Why doesn't open() return an error when a pipe open fails?

If the second argument to a piped open() contains shell
metacharacters, perl fork()s, then exec()s a shell to decode the
metacharacters and eventually run the desired program.  If the program
couldn't be run, it's the shell that gets the message, not Perl. All
your Perl program can find out is whether the shell itself could be
successfully started.  You can still capture the shell's STDERR and
check it for error messages.  See L<"How can I capture STDERR from an
external command?"> elsewhere in this document, or use the
IPC::Open3 module.

If there are no shell metacharacters in the argument of open(), Perl
runs the command directly, without using the shell, and can correctly
report whether the command started.

=head2 What's wrong with using backticks in a void context?

Strictly speaking, nothing.  Stylistically speaking, it's not a good
way to write maintainable code.  Perl has several operators for
running external commands.  Backticks are one; they collect the output
from the command for use in your program.  The C<system> function is
another; it doesn't do this.

Writing backticks in your program sends a clear message to the readers
of your code that you wanted to collect the output of the command.
Why send a clear message that isn't true?

Consider this line:

        `cat /etc/termcap`;

You forgot to check C<$?> to see whether the program even ran
correctly.  Even if you wrote

        print `cat /etc/termcap`;

this code could and probably should be written as

        system("cat /etc/termcap") == 0
        or die "cat program failed!";

which will echo the cat command's output as it is generated, instead
of waiting until the program has completed to print it out. It also
checks the return value.

C<system> also provides direct control over whether shell wildcard
processing may take place, whereas backticks do not.

=head2 How can I call backticks without shell processing?

This is a bit tricky.  You can't simply write the command
like this:

        @ok = `grep @opts '$search_string' @filenames`;

As of Perl 5.8.0, you can use C<open()> with multiple arguments.
Just like the list forms of C<system()> and C<exec()>, no shell
escapes happen.

        open( GREP, "-|", 'grep', @opts, $search_string, @filenames );
        chomp(@ok = <GREP>);
        close GREP;

You can also:

        my @ok = ();
        if (open(GREP, "-|")) {
                while (<GREP>) {
                        chomp;
                        push(@ok, $_);
                }
                close GREP;
        } else {
                exec 'grep', @opts, $search_string, @filenames;
        }

Just as with C<system()>, no shell escapes happen when you C<exec()> a
list. Further examples of this can be found in L<perlipc/"Safe Pipe
Opens">.

Note that if you're using Windows, no solution to this vexing issue is
even possible.  Even if Perl were to emulate C<fork()>, you'd still be
stuck, because Windows does not have an argc/argv-style API.

=head2 Why can't my script read from STDIN after I gave it EOF (^D on Unix, ^Z on MS-DOS)?

Some stdio's set error and eof flags that need clearing.  The
POSIX module defines clearerr() that you can use.  That is the
technically correct way to do it.  Here are some less reliable
workarounds:

=over 4

=item 1

Try keeping around the seekpointer and go there, like this:

        $where = tell(LOG);
        seek(LOG, $where, 0);

=item 2

If that doesn't work, try seeking to a different part of the file and
then back.

=item 3

If that doesn't work, try seeking to a different part of
the file, reading something, and then seeking back.

=item 4

If that doesn't work, give up on your stdio package and use sysread.

=back

=head2 How can I convert my shell script to perl?

Learn Perl and rewrite it.  Seriously, there's no simple converter.
Things that are awkward to do in the shell are easy to do in Perl, and
this very awkwardness is what would make a shell->perl converter
nigh-on impossible to write.  By rewriting it, you'll think about what
you're really trying to do, and hopefully will escape the shell's
pipeline datastream paradigm, which while convenient for some matters,
causes many inefficiencies.

=head2 Can I use perl to run a telnet or ftp session?

Try the Net::FTP, TCP::Client, and Net::Telnet modules (available from
CPAN).  http://www.cpan.org/scripts/netstuff/telnet.emul.shar
will also help for emulating the telnet protocol, but Net::Telnet is
quite probably easier to use..

If all you want to do is pretend to be telnet but don't need
the initial telnet handshaking, then the standard dual-process
approach will suffice:

        use IO::Socket;             # new in 5.004
        $handle = IO::Socket::INET->new('www.perl.com:80')
            or die "can't connect to port 80 on www.perl.com: $!";
        $handle->autoflush(1);
        if (fork()) {               # XXX: undef means failure
            select($handle);
            print while <STDIN>;    # everything from stdin to socket
        } else {
            print while <$handle>;  # everything from socket to stdout
        }
        close $handle;
        exit;

=head2 How can I write expect in Perl?

Once upon a time, there was a library called chat2.pl (part of the
standard perl distribution), which never really got finished.  If you
find it somewhere, I<don't use it>.  These days, your best bet is to
look at the Expect module available from CPAN, which also requires two
other modules from CPAN, IO::Pty and IO::Stty.

=head2 Is there a way to hide perl's command line from programs such as "ps"?

First of all note that if you're doing this for security reasons (to
avoid people seeing passwords, for example) then you should rewrite
your program so that critical information is never given as an
argument.  Hiding the arguments won't make your program completely
secure.

To actually alter the visible command line, you can assign to the
variable $0 as documented in L<perlvar>.  This won't work on all
operating systems, though.  Daemon programs like sendmail place their
state there, as in:

        $0 = "orcus [accepting connections]";

=head2 I {changed directory, modified my environment} in a perl script.  How come the change disappeared when I exited the script?  How do I get my changes to be visible?

=over 4

=item Unix

In the strictest sense, it can't be done--the script executes as a
different process from the shell it was started from.  Changes to a
process are not reflected in its parent--only in any children
created after the change.  There is shell magic that may allow you to
fake it by eval()ing the script's output in your shell; check out the
comp.unix.questions FAQ for details.

=back

=head2 How do I close a process's filehandle without waiting for it to complete?

Assuming your system supports such things, just send an appropriate signal
to the process (see L<perlfunc/"kill">).  It's common to first send a TERM
signal, wait a little bit, and then send a KILL signal to finish it off.

=head2 How do I fork a daemon process?

If by daemon process you mean one that's detached (disassociated from
its tty), then the following process is reported to work on most
Unixish systems.  Non-Unix users should check their Your_OS::Process
module for other solutions.

=over 4

=item *

Open /dev/tty and use the TIOCNOTTY ioctl on it.  See L<tty>
for details.  Or better yet, you can just use the POSIX::setsid()
function, so you don't have to worry about process groups.

=item *

Change directory to /

=item *

Reopen STDIN, STDOUT, and STDERR so they're not connected to the old
tty.

=item *

Background yourself like this:

        fork && exit;

=back

The Proc::Daemon module, available from CPAN, provides a function to
perform these actions for you.

=head2 How do I find out if I'm running interactively or not?

Good question. Sometimes C<-t STDIN> and C<-t STDOUT> can give clues,
sometimes not.

        if (-t STDIN && -t STDOUT) {
                print "Now what? ";
                }

On POSIX systems, you can test whether your own process group matches
the current process group of your controlling terminal as follows:

        use POSIX qw/getpgrp tcgetpgrp/;

        # Some POSIX systems, such as Linux, can be
        # without a /dev/tty at boot time.
        if (!open(TTY, "/dev/tty")) {
                print "no tty\n";
        } else {
                $tpgrp = tcgetpgrp(fileno(*TTY));
                $pgrp = getpgrp();
                if ($tpgrp == $pgrp) {
                        print "foreground\n";
                } else {
                        print "background\n";
                }
        }

=head2 How do I timeout a slow event?

Use the alarm() function, probably in conjunction with a signal
handler, as documented in L<perlipc/"Signals"> and the section on
"Signals" in the Camel.  You may instead use the more flexible
Sys::AlarmCall module available from CPAN.

The alarm() function is not implemented on all versions of Windows.
Check the documentation for your specific version of Perl.

=head2 How do I set CPU limits?
X<BSD::Resource> X<limit> X<CPU>

(contributed by Xho)

Use the C<BSD::Resource> module from CPAN. As an example:

        use BSD::Resource;
        setrlimit(RLIMIT_CPU,10,20) or die $!;

This sets the soft and hard limits to 10 and 20 seconds, respectively.
After 10 seconds of time spent running on the CPU (not "wall" time),
the process will be sent a signal (XCPU on some systems) which, if not
trapped, will cause the process to terminate.  If that signal is
trapped, then after 10 more seconds (20 seconds in total) the process
will be killed with a non-trappable signal.

See the C<BSD::Resource> and your systems documentation for the gory
details.

=head2 How do I avoid zombies on a Unix system?

Use the reaper code from L<perlipc/"Signals"> to call wait() when a
SIGCHLD is received, or else use the double-fork technique described
in L<perlfaq8/"How do I start a process in the background?">.

=head2 How do I use an SQL database?

The DBI module provides an abstract interface to most database
servers and types, including Oracle, DB2, Sybase, mysql, Postgresql,
ODBC, and flat files.  The DBI module accesses each database type
through a database driver, or DBD.  You can see a complete list of
available drivers on CPAN: http://www.cpan.org/modules/by-module/DBD/ .
You can read more about DBI on http://dbi.perl.org .

Other modules provide more specific access: Win32::ODBC, Alzabo, iodbc,
and others found on CPAN Search: http://search.cpan.org .

=head2 How do I make a system() exit on control-C?

You can't.  You need to imitate the system() call (see L<perlipc> for
sample code) and then have a signal handler for the INT signal that
passes the signal on to the subprocess.  Or you can check for it:

        $rc = system($cmd);
        if ($rc & 127) { die "signal death" }

=head2 How do I open a file without blocking?

If you're lucky enough to be using a system that supports
non-blocking reads (most Unixish systems do), you need only to use the
O_NDELAY or O_NONBLOCK flag from the Fcntl module in conjunction with
sysopen():

        use Fcntl;
        sysopen(FH, "/foo/somefile", O_WRONLY|O_NDELAY|O_CREAT, 0644)
                or die "can't open /foo/somefile: $!":

=head2 How do I tell the difference between errors from the shell and perl?

(answer contributed by brian d foy)

When you run a Perl script, something else is running the script for you,
and that something else may output error messages.  The script might
emit its own warnings and error messages.  Most of the time you cannot
tell who said what.

You probably cannot fix the thing that runs perl, but you can change how
perl outputs its warnings by defining a custom warning and die functions.

Consider this script, which has an error you may not notice immediately.

        #!/usr/locl/bin/perl

        print "Hello World\n";

I get an error when I run this from my shell (which happens to be
bash).  That may look like perl forgot it has a print() function,
but my shebang line is not the path to perl, so the shell runs the
script, and I get the error.

        $ ./test
        ./test: line 3: print: command not found

A quick and dirty fix involves a little bit of code, but this may be all
you need to figure out the problem.

        #!/usr/bin/perl -w

        BEGIN {
        $SIG{__WARN__} = sub{ print STDERR "Perl: ", @_; };
        $SIG{__DIE__}  = sub{ print STDERR "Perl: ", @_; exit 1};
        }

        $a = 1 + undef;
        $x / 0;
        __END__

The perl message comes out with "Perl" in front.  The BEGIN block
works at compile time so all of the compilation errors and warnings
get the "Perl:" prefix too.

        Perl: Useless use of division (/) in void context at ./test line 9.
        Perl: Name "main::a" used only once: possible typo at ./test line 8.
        Perl: Name "main::x" used only once: possible typo at ./test line 9.
        Perl: Use of uninitialized value in addition (+) at ./test line 8.
        Perl: Use of uninitialized value in division (/) at ./test line 9.
        Perl: Illegal division by zero at ./test line 9.
        Perl: Illegal division by zero at -e line 3.

If I don't see that "Perl:", it's not from perl.

You could also just know all the perl errors, and although there are
some people who may know all of them, you probably don't.  However, they
all should be in the perldiag manpage. If you don't find the error in
there, it probably isn't a perl error.

Looking up every message is not the easiest way, so let perl to do it
for you.  Use the diagnostics pragma with turns perl's normal messages
into longer discussions on the topic.

        use diagnostics;

If you don't get a paragraph or two of expanded discussion, it
might not be perl's message.

=head2 How do I install a module from CPAN?

The easiest way is to have a module also named CPAN do it for you.
This module comes with perl version 5.004 and later.

        $ perl -MCPAN -e shell

        cpan shell -- CPAN exploration and modules installation (v1.59_54)
        ReadLine support enabled

        cpan> install Some::Module

To manually install the CPAN module, or any well-behaved CPAN module
for that matter, follow these steps:

=over 4

=item 1

Unpack the source into a temporary area.

=item 2

        perl Makefile.PL

=item 3

        make

=item 4

        make test

=item 5

        make install

=back

If your version of perl is compiled without dynamic loading, then you
just need to replace step 3 (B<make>) with B<make perl> and you will
get a new F<perl> binary with your extension linked in.

See L<ExtUtils::MakeMaker> for more details on building extensions.
See also the next question, "What's the difference between require
and use?".

=head2 What's the difference between require and use?

(contributed by brian d foy)

Perl runs C<require> statement at run-time. Once Perl loads, compiles,
and runs the file, it doesn't do anything else. The C<use> statement
is the same as a C<require> run at compile-time, but Perl also calls the
C<import> method for the loaded package. These two are the same:

        use MODULE qw(import list);

        BEGIN {
                require MODULE;
                MODULE->import(import list);
                }

However, you can suppress the C<import> by using an explicit, empty
import list. Both of these still happen at compile-time:

        use MODULE ();

        BEGIN {
                require MODULE;
                }

Since C<use> will also call the C<import> method, the actual value
for C<MODULE> must be a bareword. That is, C<use> cannot load files
by name, although C<require> can:

        require "$ENV{HOME}/lib/Foo.pm"; # no @INC searching!

See the entry for C<use> in L<perlfunc> for more details.

=head2 How do I keep my own module/library directory?

When you build modules, tell Perl where to install the modules.

For C<Makefile.PL>-based distributions, use the INSTALL_BASE option
when generating Makefiles:

        perl Makefile.PL INSTALL_BASE=/mydir/perl

You can set this in your CPAN.pm configuration so modules automatically install
in your private library directory when you use the CPAN.pm shell:

        % cpan
        cpan> o conf makepl_arg INSTALL_BASE=/mydir/perl
        cpan> o conf commit

For C<Build.PL>-based distributions, use the --install_base option:

        perl Build.PL --install_base /mydir/perl

You can configure CPAN.pm to automatically use this option too:

        % cpan
        cpan> o conf mbuild_arg --install_base /mydir/perl
        cpan> o conf commit

INSTALL_BASE tells these tools to put your modules into
F</mydir/perl/lib/perl5>.  See L<How do I add a directory to my
include path (@INC) at runtime?> for details on how to run your newly
installed moudles.

There is one caveat with INSTALL_BASE, though, since it acts
differently than the PREFIX and LIB settings that older versions of
ExtUtils::MakeMaker advocated. INSTALL_BASE does not support
installing modules for multiple versions of Perl or different
architectures under the same directory. You should consider if you
really want that , and if you do, use the older PREFIX and LIB
settings. See the ExtUtils::Makemaker documentation for more details.

=head2 How do I add the directory my program lives in to the module/library search path?

(contributed by brian d foy)

If you know the directory already, you can add it to C<@INC> as you would
for any other directory. You might <use lib> if you know the directory
at compile time:

        use lib $directory;

The trick in this task is to find the directory. Before your script does
anything else (such as a C<chdir>), you can get the current working
directory with the C<Cwd> module, which comes with Perl:

        BEGIN {
                use Cwd;
                our $directory = cwd;
                }

        use lib $directory;

You can do a similar thing with the value of C<$0>, which holds the
script name. That might hold a relative path, but C<rel2abs> can turn
it into an absolute path. Once you have the

        BEGIN {
                use File::Spec::Functions qw(rel2abs);
                use File::Basename qw(dirname);

                my $path   = rel2abs( $0 );
                our $directory = dirname( $path );
                }

        use lib $directory;

The C<FindBin> module, which comes with Perl, might work. It finds the
directory of the currently running script and puts it in C<$Bin>, which
you can then use to construct the right library path:

        use FindBin qw($Bin);

=head2 How do I add a directory to my include path (@INC) at runtime?

Here are the suggested ways of modifying your include path, including
environment variables, run-time switches, and in-code statements:

=over 4

=item the PERLLIB environment variable

        $ export PERLLIB=/path/to/my/dir
        $ perl program.pl

=item the PERL5LIB environment variable

        $ export PERL5LIB=/path/to/my/dir
        $ perl program.pl

=item the perl -Idir command line flag

        $ perl -I/path/to/my/dir program.pl

=item the use lib pragma:

        use lib "$ENV{HOME}/myown_perllib";

=back

The last is particularly useful because it knows about machine
dependent architectures.  The lib.pm pragmatic module was first
included with the 5.002 release of Perl.

=head2 What is socket.ph and where do I get it?

It's a Perl 4 style file defining values for system networking
constants.  Sometimes it is built using h2ph when Perl is installed,
but other times it is not.  Modern programs C<use Socket;> instead.

=head1 REVISION

Revision: $Revision$

Date: $Date$

See L<perlfaq> for source control details and availability.

=head1 AUTHOR AND COPYRIGHT

Copyright (c) 1997-2009 Tom Christiansen, Nathan Torkington, and
other authors as noted. All rights reserved.

This documentation is free; you can redistribute it and/or modify it
under the same terms as Perl itself.

Irrespective of its distribution, all code examples in this file
are hereby placed into the public domain.  You are permitted and
encouraged to use this code in your own programs for fun
or for profit as you see fit.  A simple comment in the code giving
credit would be courteous but is not required.