I would say in general, for reasons that are independent of performance, a lexical "my" variable is preferred over a package level "our" variable. Also besides scoping issues, a "my" variable is faster than a "local" variable. A shift is very slightly faster than a "my $var = @_;"( missed parens! oops..(my $var)= @_;) when the sub has one arg passed to it. But in this case, I don't see how it could make much difference.

If you are going to do some benchmarking, then I would suggest this code:

print_array2 (\@foo);

sub print_array2 
{
   my $array_ref = shift;
   print join ("\n", @$array_ref), "\n";
}
[download]

I am curious as what you are doing as I've never seen a Perl program that was limited by speed of output (much more problematic is processing of the input data!). Print like all I/O is an expensive operation and time to do it usually dwarfs any calculation required to get the thing to print formatted. I would at least suggest adding the above code into your benchmarking - it will run very quickly.Of course, you can also try:

sub print_array2 
{
   my $array_ref = shift;
   foreach (@$array_ref))
   {
      print "$_\n";
   }
}
[download]