And here's the Forking::Amazing version. See Re: figuring out Parallel::ForkManager slots for the Forking::Amazing module, which I wrote because I didn't particularly care for Parallel::ForkManager. I think Forking::Amazing makes it clearer what runs in the child and what runs in the parent, and makes it easier to pass results from the child back to the parent.

#!/usr/bin/perl

use strict; # https://perlmonks.org/?node_id=11150996
use warnings;
use Forking::Amazing;

my @testLists = 1 .. 10; # FAKE for testing
my $opts = 'opts'; # FAKE for testing
my $status = 0;

Forking::Amazing::run
  2,                                          # max forks
  sub { return [ regressions($opts, @_) ]; }, # child
  sub { $status += $_[1][0]; },               # parent
  @testLists;                                 # argument for each fork
print "status $status\n";

sub regressions # FIXME for testing
  {
  print time, " pid $$ starting on (@_)\n";
  select undef, undef, undef, 1 + rand 3;
  print time, " pid $$ ending on (@_)\n";
  return pop;
  }
[download]