When fork() is called, you end up having two processes: the parent and the child. They both execute the same code (your Perl script), but you can understand if you're in the child testing whether the call returned 0. Then, you're supposed to call exec() (if you really need it) passing the control to the process you want to execute, but this isn't necessary if the code is in the parent script!

The two processes created by fork() are independent, so they will be both running after the call, if this is what you're worried about. Unlike system(), which waits for the child process to complete, after fork() you can do whatever you want in the parent process, even calling fork() two more times to complete the creation of your three subprocesses.

Simple snippet:

#!/usr/bin/perl
use strict;
use warnings;
sub child {
   print("Hey, I'm child $$\n");
   sleep(2 + rand(5));
   print("$$ exiting...\n");
   exit(0);
}

foreach (1 .. 3) {
   my $pid = fork;
   die "fork(): $@, stopped" unless defined($pid);
   child() unless $pid;
}

# Only the parent reaches this point
wait foreach (1 .. 3);
[download]

Hey, I'm child 1048
Hey, I'm child 1368
Hey, I'm child 424
1368 exiting...
424 exiting...
1048 exiting...
[download]

Thanks for your time and sorry if I am getting you confused. What I basically need is to make 3 system type calls to run 3 seperate c programs at once from a master perl script and then ammalgumate the ouput from the 3. The reason for having concurrent jobs is to trick openmosix into loadbalance the jobs on a cluster. I was under the impression that perl fork would wait to finish the child process as it would with a system call.

Thanks for your help. You have cleared the confusion for me. I have managed to get it working as I want it to. Many thanks.