Update: removed unnecessary intermediate array

my @sorted;
for my $key (keys %entry) {
    my $scores = $entry{$key}{'Score'};
    my $locations = $entry{$key}{'Location'};
    my @array = map { [ $key, $scores->[$_],$locations->[$_] ] } 0 .. 
+$#$scores;
    @array = sort { $b->[1] <=> $a->[1] } @array;
    push(@sorted, [ $array[0][1], \@array ]);
}
@sorted = sort { $b->[0] <=> $a->[0] } @sorted;

for my $array (@sorted) {
    my $aoa = $array->[1];
    for my $result (@$aoa) {
        print "$result->[0] $result->[1] $result->[2]\n";
    }
}
[download]

TRA 23 4-19
TRA 15 7-15
TRA 2 78-120
BLA 10 2-10
BLA 5 1-10
[download]

Brilliant! Exactly what I wanted. Thanks a lot.

Nope. Hashes can't be sorted, only their keys can be sorted and held in a separate array.

Otherwise please elaborate better what you want to achieve your post is rather cryptic...

I see thanks for that. I have a list of IDs with a corresponding score and location values. I want to order the IDs based on highest score associated with them. I also want the IDs to be grouped together. When I use array of hashes like below how would I order it then?

push @entry, { 
    ID                 => 'BLA',
    Score             => 5,
    Location         => '1-10',
};

push @entry, { 
    ID                 => 'TRA',
    Score             => 15,
    Location         => '7-15',
};

push @entry, { 
    ID                 => 'BLA',
    Score             => 10,
    Location         => '2-10',
};

Output; 

TRA    15 7-15
BLA 10 2-10
BLA 5, 1-10
[download]

That's a completely different data structure now!!!

 @sorted = sort { $a->{Score} <=> $b->{Score} } @entry

Does it.

Cheers Rolf

(addicted to the Perl Programming Language)

#!perl
use strict;    
use warnings;
use DBI;

# set up
my %entry=();
while (<DATA>){
  my ($id,$score,$locn) = split '\s+',$_;
  push @{$entry{$id}{Score}}   ,$score;
  push @{$entry{$id}{Location}},$locn;
}

# create db
my $dbfile = 'scores.sqlite';
unlink($dbfile);
my $dbh = DBI->connect('dbi:SQLite:dbname='.$dbfile , undef , undef , 
                {RaiseError =>1, AutoCommit =>1}) or die $DBI::errstr;
$dbh->do('CREATE TABLE SCORES (ID,SCORE integer,LOCATION)');
$dbh->do('CREATE TABLE LEAGUE (ID, HI_SCORE integer)');

# load data
my $sth = $dbh->prepare('INSERT INTO SCORES VALUES (?,?,?)');
for my $id (keys %entry){
  my $rec = $entry{$id};
  my @score = @{$rec->{'Score'}};
  my @locn  = @{$rec->{'Location'}};
  for my $i (0..$#score){
    $sth->execute($id,$score[$i],$locn[$i]);
  }
}

# aggregate
$dbh->do('INSERT INTO LEAGUE (ID, HI_SCORE)
          SELECT ID, MAX(SCORE) FROM SCORES
          GROUP BY ID');
# report
my $sql = 'SELECT S.ID, SCORE, LOCATION
           FROM SCORES AS S
           LEFT JOIN LEAGUE AS L
           ON S.ID = L.ID
           ORDER BY HI_SCORE DESC, SCORE DESC';
        
my $ar = $dbh->selectall_arrayref($sql);
for (@$ar){
  printf "%-5s %5d %10s\n",@$_,"\n";
}
$dbh->disconnect;

__DATA__
TRA 15 7-15
TRA 23 4-19
TRA 2  78-120
BLA 5  1-10
BLA 10 2-10
[download]

Interesting idea, will give it a try. Thanks for that

Here's a decorate-sort-undecorate (GRT) variation that may be useful if speed is an issue. Speaking of issues, there is a corner-case not handled here in which a  Score array (and its corresponding  Location array) is empty. If you're creating entries by push-ing to a hash as shown in the OP, I don't think this situation can arise, but I just thought I'd mention it.

Will have a look how much faster it is. Thanks for the input.