NAME

System2 - like system(), but with access to STDOUT and STDERR.

SYNOPSIS

use System2;

$System2::debug++;

my ($out, $err) = system2(@args);

my ($exit_value, $signal_num, $dumped_core) = &System2::exit_status($?);

print "EXIT: exit_value $exit_value signal_num ".
      "$signal_num dumped_core $dumped_core\n";

print "OUT:\n$out";
print "ERR:\n$err"

DESCRIPTION

The module presents an interface for executing a command, and gathering the output from STDOUT and STDERR.

Benefits of this interface:

-

the Bourne shell is never implicitly invoked: saves a stray exec(), and bypasses those nasty shell quoting problems.

-

cheaper to run than open3().

-

augmented processing of arguments, to allow for overriding arg[0] (eg. initiating a login shell).

STDOUT and STDERR are returned in scalars. $? is set. (Split on $/ if you want the expected lines back.)

If $debug is set, on-the fly diagnostics will be reported about how much data is being read.

Provides for convenience, a routine exit_status() to break out the exit value into separate scalars, straight from perlvar(1):

-

the exit value of the subprocess

-

which signal, if any, the process died from

-

reports whether there was a core dump.

There are two interfaces available: a regular list, or named parameters:

These are equivalent:

my @args = ( '/bin/sh', '-x', '-c', 'echo $0' );

my @args = ( path => '/bin/sh', args => [ '-c', 'echo $0' ] );

To override arg[0], pass in a arrayref for the first argument, or use the arg0 named parameter. Contrast the prior argument lists with these below:

my @args = ( ['/bin/sh', '-sh'], '-c', 'echo $0' );

my @args = ( path => '/bin/sh', args => ['-c', 'echo $0'],
             arg0 => '-sh' );

CAVEATS

Obviously, the returned scalars can be quite large, depending on the nature of the program being run. In the future, I intend to introduce options to allow for temporary file handles, but for now, be aware of the potential resource usage.

Although I've been using this module for literally years now personally, consider it lightly tested, until I get feedback from the public at large. (Treat this as a hint to tell me that you're using it. :)

Have at it.

AUTHOR

Brian Reichert <reichert@numachi.com>

SEE ALSO

perlfunc(1), perlvar(1).