Going back to my attitude when I was a C programmer: It pays to know how your compiler thinks. (Needs adjustment for application to modern use of Perl, but the sentiment is the same.)

Adding/replacing these three lines into my original script above:

use Encode qw( encode_utf8 );
my $x = chr(1 << 63);
my @Unsorted = ( 'Dog', 'Cat', 'Bird', undef, $x, 'Elephant', undef, '
+Lizard' );
[download]

Yields:

S:\Steve\Dev\PerlMonks\P-2017-06-12@0734-sort-undef>perl .\sort011.pl
----------------------------------------------------------------------
+---------
Original:
----------------------------------------------------------------------
+---------
Dog
Cat
Bird
(undef)
Wide character in print at .\sort011.pl line 58.
 ÇêÇÇÇÇÇÇÇÇÇÇ
Elephant
(undef)
Lizard
----------------------------------------------------------------------
+---------
----------------------------------------------------------------------
+---------
Custom Sort:
----------------------------------------------------------------------
+---------
Bird
Cat
Dog
Elephant
Lizard
(undef)
(undef)
Wide character in print at .\sort011.pl line 58.
 ÇêÇÇÇÇÇÇÇÇÇÇ
----------------------------------------------------------------------
+---------

S:\Steve\Dev\PerlMonks\P-2017-06-12@0734-sort-undef>
[download]

A string of chr(255)bytes longer than the longest item in the original array still fails to sort to the bottom; knowing that Unicode characters are stored differently than old-fashioned ASCII strings empowers the Perl programmer to make a better choice.

Thank you for the information!

I'd upvote the post, but there isn't any point, as it's Anonymous Monk.