I've been watching this thread for the last day or so. The problem with your code has been identified. You've received a number of working solutions, all of which I've upvoted.

Having said that, all solutions here seem more complicated than is necessary. To my mind, the following is really all you need:

$ perl -e '
    my %seen;
    my @values = grep !$seen{$_}++, @ARGV;
    print "@values\n";
' 1 2 1 2 3 1 2 3
1 2 3
[download]

There are a number of ways this can be shortened. Here are a couple of examples:

$ perl -E '
    local $, = " ";
    say grep {state $seen; !$seen->{$_}++} @ARGV;
' 1 2 1 2 3 1 2 3
1 2 3
[download]

$ perl -E 'say "@{[grep {state $seen; !$seen->{$_}++} @ARGV]}";' 1 2 1
+ 2 3 1 2 3
1 2 3
[download]

However, there are potential drawbacks to these:

• They involve cleverness. In other words, you, or other maintainers, may not immediately understand the code on subsequent inspection. This can reduce maintainability — even if that just involves the time needed for research. For instance, at first glance, without referring to doco, do you know what these mean:
  • $,
  • @{[...]}
• You'll need Perl v5.10 or later.
• If this code exists in a loop with many iterations, or an often called subroutine, you may start to see efficiency degradation. grep has two forms, grep BLOCK LIST & grep EXPR,LIST. The BLOCK form is slower. Use Benchmark if this is important in your code.