I'm a fan of the GRT sort. I like it because it allows multiple keys to be sorted simultaneously in one action. Whether this is a narrow-minded view I have yet to learn... (which I'm still doing :p)

Full info if your interested: Advanced Sorting - GRT - Guttman Rosler Transform
Original whitepaper discussing sorting in general and GRT: A Fresh Look at Efficient Perl Sorting

The basic structure of the grt is:

map {
  unpack or substr
}sort map{
  pack [template, $var1, $var2, etc, $string_or_reference]
}
[download]

with the main principle being to let Perl use its highly optimised default sorting algorithm. When tweaking how the sort function works, it can get very slow and inefficient pretty easily. With the GRT sort you call the sort function either as plain old sort or reverse sort

If you really don't care about how sorting works, take a look at Sort::Maker. It takes a series of arguments and creates a sorting sub for you based on the criteria you provide.

If you are interested on how a GRT sort could be applied to your problem keep reading. :) (and this is by no means the one and only solution. I'm actually interested to see other monks comments, and if my below solution could be improved further)

#! usr/bin/perl -w
use strict;
use warnings;

my @original_array = <DATA>;

foreach (@original_array){
    print "$_";
}

my @sorted_array = 
    map{
        substr($_,68);
        #unpack('x68 A*', $_)    #achieves same thing but apparently l
+ess efficient...
    } sort map {
        my ($perm, $path, $filename) = split /\s+/, $_;
        pack 'N A32 A32 A*', $perm, $path, $filename, $_;
} (@original_array);
                            
print "\n\n";

foreach (@sorted_array){
    print "$_";
}




__DATA__
2755    home
444    home/backup    appletest.txt
444    home/backup    dhl.txt
444    home/support    appletest.bat
2755    bin
755    bin/backup    env.txt
755    bin/support    arc.bat
755    bin/backup    aus.txt
2755     etc    
644    etc/backup    appletest.txt
644    etc/support    arc.bat
644    etc/support    dhl.bat
644    etc/support    env.bat
[download]

From what I've learnt so far, the pack is essentially prepending a header on the front of each line of text, which the sort function only ever reads upto, before deciding whether two lines are le, ge, or eq.

The substr (and unpack), essentially strip off the header portion and return into the array @sorted_array the original line of text.

Hope this is helpful to you