The London Perl and Raku Workshop takes place on 26th Oct 2024. If your company depends on Perl, please consider sponsoring and/or attending.


Proc::Fork - simple, intuitive interface to the fork() system call


 use Proc::Fork;

 run_fork {
     child {
         # child code goes here.
     parent {
         my $child_pid = shift;
         # parent code goes here.
         waitpid $child_pid, 0;
     retry {
         my $attempts = shift;
         # what to do if fork() fails:
         # return true to try again, false to abort
         return if $attempts > 5;
         sleep 1, return 1;
     error {
         # Error-handling code goes here
         # (fork() failed and the retry block returned false)


This module provides an intuitive, Perl-ish way to write forking programs by letting you use blocks to illustrate which code section executes in which fork. The code for the parent, child, retry handler and error handler are grouped together in a "fork block". The clauses may appear in any order, but they must be consecutive (without any other statements in between).

All four clauses need not be specified. If the retry clause is omitted, only one fork will be attempted. If the error clause is omitted the program will die with a simple message if it can't retry. If the parent or child clause is omitted, the respective (parent or child) process will start execution after the final clause. So if one or the other only has to do some simple action, you need only specify that one. For example:

 # spawn off a child process to do some simple processing
 run_fork { child {
     exec '/bin/ls', '-l';
     die "Couldn't exec ls: $!\n";
 } };
 # Parent will continue execution from here
 # ...

If the code in any of the clauses does not die or exit, it will continue execution after the fork block.


All of the following functions are exported by default:


 run_fork { ... }

Performs the fork operation configured in its block.


 child { ... }

Declares the block that should run in the child process.


 parent { ... }

Declares the block that should run in the parent process. The child's PID is passed as an argument to the block.


 retry { ... }

Declares the block that should run in case of an error, ie. if fork returned undef. If the code returns true, another fork is attempted. The number of fork attempts so far is passed as an argument to the block.

This can be used to implement a wait-and-retry logic that may be essential for some applications like daemons.

If a retry clause is not used, no retries will be attempted and a fork failure will immediately lead to the error clause being called.


 error { ... }

Declares the block that should run if there was an error, ie when fork returns undef and the retry clause returns false. The number of forks attempted is passed as an argument to the block.

If an error clause is not used, errors will raise an exception using die.


The distribution includes the following examples as separate files in the eg/ directory:

Simple example with IPC via pipe

 use strict;
 use Proc::Fork;

 use IO::Pipe;
 my $p = IO::Pipe->new;

 run_fork {
     parent {
         my $child = shift;
         print while <$p>;
         waitpid $child,0;
     child {
         print $p "Line 1\n";
         print $p "Line 2\n";
     retry {
         if( $_[0] < 5 ) {
             sleep 1;
             return 1;
         return 0;
     error {
         die "That's all folks\n";

Multi-child example

 use strict;
 use Proc::Fork;
 use IO::Pipe;

 my $num_children = 5;    # How many children we'll create
 my @children;            # Store connections to them
 $SIG{CHLD} = 'IGNORE';   # Don't worry about reaping zombies

 # Spawn off some children
 for my $num ( 1 .. $num_children ) {
     # Create a pipe for parent-child communication
     my $pipe = IO::Pipe->new;

     # Child simply echoes data it receives, until EOF
     run_fork { child {
         my $data;
         while ( $data = <$pipe> ) {
             chomp $data;
             print STDERR "child $num: [$data]\n";
     } };

     # Parent here
     push @children, $pipe;

 # Send some data to the kids
 for ( 1 .. 20 ) {
     # pick a child at random
     my $num = int rand $num_children;
     my $child = $children[$num];
     print $child "Hey there.\n";

Daemon example

 use strict;
 use Proc::Fork;
 use POSIX;

 # One-stop shopping: fork, die on error, parent process exits.
 run_fork { parent { exit } };

 # Other daemon initialization activities.
 $SIG{INT} = $SIG{TERM} = $SIG{HUP} = $SIG{PIPE} = \&some_signal_handler;
 POSIX::setsid() == -1 and die "Cannot start a new session: $!\n";
 close $_ for *STDIN, *STDOUT, *STDERR;

 # rest of daemon program follows

Forking socket-based network server example

 use strict;
 use IO::Socket::INET;
 use Proc::Fork;


 my $server = IO::Socket::INET->new(
     LocalPort => 7111,
     Type      => SOCK_STREAM,
     Reuse     => 1,
     Listen    => 10,
 ) or die "Couln't start server: $!\n";

 my $client;
 while ($client = $server->accept) {
     run_fork { child {
         # Service the socket
         print $client "Ooga! ", time % 1000, "\n";
         exit; # child exits. Parent loops to accept another connection.
     } }


Aristotle Pagaltzis <>

Documentation by Eric J. Roode.


This software is copyright (c) 2018 by Aristotle Pagaltzis. Its documentation is copyright (c) 2002 by Eric J. Roode.

This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.