use 5.012;
use warnings;

use constant {
    # Common collation sequences
    ALPHA => sub { my ($a, $b) = @_; $a cmp $b },
    NUMERIC => sub { my ($a, $b) = @_; $a <=> $b },
    # Sort orders
    ASC => +1,
    DSC => -1
};

# Multicolumn sort function.
# Params:
#   $columns = Arrayref of sort column definitions in priority order.
#              Each column definition is a hashref with the keys
#                index   => Column index in the data rows
#                collate => Binary subref defining collation order
#                order   => ASC for ascending or DSC for descending order
#   $data    = Arrayref of data rows
# Returns:
#   Arrayref of sorted data rows
sub multisort {
    my ($columns, $data) = @_;
    [sort {
        for my $column (@$columns) {
            my $d = $column->{order} * $column->{collate}->($a->[$column->{index}], $b->[$column->{index}]);
            return $d unless $d == 0;
        }
    } @$data];
}

# Runnable example:
my @data = (
    ['aaron', 'zander', 20],
    ['hannes', 'zander', 20],
    ['aaron', 'suppe', 10],
);

my @columns = (
    {index => 2, collate => NUMERIC, order => ASC},
    {index => 1, collate => ALPHA, order => ASC},
    {index => 0, collate => ALPHA, order => ASC},
);

use Data::Dumper;
print Dumper multisort(\@columns, \@data);