Some minor (possible) improvements:

sub next_prime_iterator {
   my $num = shift(@_);
   return 3   if  $num < 3;
   do { $num += 2 } until is_prime($num);
   return $num;
}

sub is_prime {
   my $guess = shift(@_);

   my $divisor = 3;
   my $quotient;
   while(1) {
       $quotient= $guess / $divisor;
       return 1   if  $quotient < $divisor;
       return 0   if  int($quotient) == $quotient;
       $divisor += 2;
   }
}
[download]

Note that using int($quotient) == $quotient made me worry that this wouldn't always work if $guess is very near the mantissa limit on your machine. But it turns out that Perl's % doesn't do any better there anyway (it does worse):

my $x= 1+~0;
$x *= 2 while $x != $x+1;
$x /= 2;
my $y= $x;
for( 1..5 ) {
    $x++;
    print $x-$y, " ", $x%3, " ", $x/3-int($x/3), "; ";
}
print $/
__END__
1 0 0.75; 2 0 0; 3 2 0.25; 4 2 0.75; 5 1 0;
[download]

So the first numbers (1 2 3 4 5) show that Perl can handle those integers exactly. The second numbers (0 0 2 2 1) don't follow the pattern of 2 0 1 2 0 1 2 0 1 as they should so % is returning invalid results here. The third numbers (0.75 0 0.25 0.75 0) are following the correct pattern (something close to 0.66 0 0.33 0.66 0 0.33) so I trust my int($quotient) == $quotient test.