#! /usr/bin/perl  -l

my $zoo = {
  Mountain => { Dragon => 'Deadly', Troll => 'Deadly', Nymph => 'Oread
+', Snake => 'Deadly' },
  Cave => { Dragon => 'Deadly', Spider => 'Poisonous', Troll => 'Deadl
+y', Ogre => 'Deadly' },
  Forest => { Elf => 'Dark', Troll => 'Deadly', Nymph => 'Alseid', Spi
+der => 'Hairy' },
  Lake => { Nymph => 'Naiad', Snake => 'Poisonous', Serpent => 'Deadly
+' },
  Sea => { Kraken => 'Monstrous', Serpent => 'Monstrous' }
};

sub each_uniq {
  my @refs = @_;
  my %seen = ();
  sub {
    while (@refs) {
      my @pair = each %{$refs[0]} or shift @refs and next;
      return @pair unless $seen{$pair[0]}++;
    }
    return;
  }
}

my $iter = each_uniq(values %$zoo);
print join(", ", $iter->()) for 1..5;
[download]

Update. I read the previous questions monkini has asked, and it would appear the problem is largely unrelated to questions asked. From what I gather, he has a number of clusters each with a number of (labelled) points, and distances between points given. The problem apparently is to list the pairs of points in a cluster sorted by their distance, for each cluster separately. If so, the solution is to map clusters to points (HoA), iterate, slice map slice sort the HoH, print.

Maybe you could provide a Data::Dumper output of your structure (or part thereof) to clarify your problem.

I agree ... I think that we really need to see this thing, in order to comment meaningfully about it and thus answer the OP question.   Nothing less will do in this case.

Hashes are unsorted in Perl, there is nothing like "second keys".

Programming is an art of speaking, maybe it helps rephrasing your ideas in proper English?

-> How do I post a question effectively?

oops sorry, I was unclear here. The hash looks like:

$hash{$key1}{$key2} = $value
[download]

The subroutine 'retrieve_kv' already retrieves the other key and corresponding value, my question is how to put it in the loop, so that the keys retrieved are unique (there can be duplicates). Basically sth like:

do {
my ($key, $val) = retrieve_kv($hash{$k});
 } until (ALL 5 KEYS ARE UNIQUE);
[download]

Hash keys are always unique, duplicates are impossible.

-> How do I post a question effectively?

Cheers Rolf

( addicted to the Perl Programming Language)

Use List::MoreUtils::uniq().

use strict;
use warnings;

use List::MoreUtils qw( uniq );

my %deck;

for my $suit (qw( Clubs Hearts Diamonds Spades )) {
    for my $rank (qw( 2 3 4 5 6 7 8 9 10 J Q K A )) {
        $deck{$suit}{$rank} = 1;
    }
}

$, = " ";
$\ = "\n";

# Prints 10 2 3 4 5 6 7 8 9 A J K Q
print sort { $a cmp $b } uniq map { keys $deck{$_} } keys %deck;
[download]

If so I'd consider a count of keys to drive the decision to dive any deeper. Carry the prefix with you as you recurse so you can assemble the full key path before returning it to the calling layer.