That's nice. Although it is slower than using a hash. Comparing your code against this routine:

sub sort_unique_hash {
    my %hash;
    @hash{@_} = ();
    return sort keys %hash;
}
[download]

gives (where "foreach" is your code and "hash" is the above):

Benchmark: timing 100000 iterations of foreach, hash...
foreach: 14 secs (13.50 usr  0.00 sys = 13.50 cpu)
hash:  6 secs ( 5.85 usr  0.00 sys =  5.85 cpu)
[download]

This is with:

@list = qw/foo bar baz foo quack bar baz/;
[download]

The difference is even more marked if you have rather odd arrays that you want to sort -u. For example, running the code against this array:

@list = ('foo') x 1000;
[download]

gives

Benchmark: timing 1000 iterations of foreach, hash...
foreach: 12 secs (11.82 usr  0.00 sys = 11.82 cpu)
hash:  1 secs ( 0.88 usr  0.00 sys =  0.88 cpu)
[download]

Presumably because your code sorts the list before it selects only the unique elements, and the hash approach only sorts the unique list--so you're sorting 1000 elements, and the hash approach just sorts 1.

Just something to think about.