view release on metacpan or  search on metacpan

lib/Acme/Spork.pm  view on Meta::CPAN

    require Proc::Daemon;
    {
        local $SIG{'HUP'} = $SIG{'HUP'} || ''; # workaround until http://rt.cpan.org/Public/Bug/Display.html?id=21453
        goto &Proc::Daemon::Init;
    }
}

sub daemonize_without_close_on {
    require Proc::Daemon;
    {
        no warnings 'redefine';
        local *POSIX::close = sub { return 1; }; # the "without_close_on" part

        local $SIG{'HUP'} = $SIG{'HUP'} || ''; # workaround until http://rt.cpan.org/Public/Bug/Display.html?id=21453
        Proc::Daemon::Init(@_);
    }
}

1;

__END__

=head1 NAME

Acme::Spork - Perl extension for spork()ing in your script

=head1 SYNOPSIS

    use Acme::Spork;
    my $spork_pid = spork(\&long_running_code, @ARGV) 
        or die "Could not fork for spork: $!";
    print "Long running code has been started as PID $spork_pid, bye!\n";

=head1 DESCRIPTION

A spork in the plastic sense is a fork combined with a spoon. In programming I've come to call a spork() a fork() that does more than just a fork.

I use it to describe when you want to fork() to run some long running code but immediately return to the main program instead of waiting for it.

=head1 spork()

The first argument is a code ref that gets executed and any other args are passed to the call to the code ref.

    #!/usr/bin/perl

    use strict;
    use warnings;
    use Acme::Spork;

    print 1;
    spork( 
        sub { 
            sleep 5;
            open my $log_fh, '>>', 'spork.log', or die "spork.log open failed: $!";
            print {$log_fh} "I am spork hear me spoon\n"; 
            close $log_fh;
        },
    ) or die qq{Couldn't fork for spork: $!};
    print 2;

This prints out "12" immediately and is done running, now if you tail -f spork.log you'll see "I am spork hear me spoon\n" get written to it 4 or 5 seconds later by the spork()ed process :)

spork() returns the PID of the spork()ed process so you can keep track of them and do what you need with it.

If it returns false then fork failed so you can:

    if(spork(\&foo)) {
        print "I am spork here me spoon\n";
    }
    else {
        print "Could not fork for spork: $!";
    }
    my $spork_pid = spork(\&foo) or die "Could not fork for for spork: $!";

=head2 %reopen_stdfhs_to

You can now have spork reopen one or more of STDIN, STDOUT, STDERR.
You define how this is handled in the Acme::Spork hash '%reopen_stdfhs_to'

The key is the STD* handle (any other values are simply ignored).

The value is an array reference of arguments to open().

Its always a good idea to local()ize it as well (and specify all 3 handles, otherwise you may get some strange warnings and behavior):

    local %Acme::Spork::reopen_stdfhs_to = (
         STDIN  => [qw(< /dev/null)],
         STDOUT => [qw(> /dev/null)],
         STDERR => [qw(> &STDOUT)],
    );
    spork(...)    
    
or you can have it set to the value above globally like so:

    use Acme::Spork qw(-reopen_stdfhs);
    ...
    spork(...)

=head2 setsid()

Say you have a custom module that is a subset of POSIX functions that includes setsid() and you'd rather use that one instead of bringing in all of POSIX.

Just define Acme::Spork::setsid ()

   sub Acme::Spork::setsid {
       require POSIX::Subset;
       POSIX::Subset::setsid();  
   }
   my $pid = spork(...);

or if you already have the module loaded:

  *Acme::Spork::setsid = *POSIX::Subset::setsid;
   my $pid = spork(...);

It will use your setsid() instead and POSIX will not be brought in. If it doesn't actually setsid() then you just broke yourself so don't do that.

=head1 daemonize()

Since many daemons need to spork a child process when a request is received I've included a cheat function to daemonize your script execution.