Hi rsiedl,

You say you want to find an "easier/quicker" way to find the highest value in a hash. Well, do you mean quicker to develop (less lines), or quicker to run (faster execution)? I'm assuming that "easier" means easier to read or understand?

Anyway, the quickest (both meanings) and easiest way is to just use the exact method you have already discared. Why? The sort built-in will always be the quickest sorting method for general numeric or string data, since it is a Perl built-in written directly in C, and is probably one of the most optimised Perl built-in functions (in fact, I wouldn't be surprised if my Benchmark timings below are inaccurate simply because of some optimisation or other I am unaware of). Also, the values function is the quickest way of getting at the values of a hash if you don't care about the keys. Combining sort and values is the quickest, and in my opinion, easiest way of achieving your objective.

my %hashName = ( "term1" => 83, "term2" => 20, "term3" => 193 );
my ($max) = sort { $b <=> $a } values %hashName;
[download]

This Benchmark script hopefully proves my point sufficiently well:

#!/usr/bin/perl

use Benchmark qw(timethese cmpthese);

my %hashName = ( "term1" => 83, "term2" => 20, "term3" => 193 );

print "Timings:\n\n";
my $r = timethese(1000000, {
    'orig' => sub {
        my $i = 0;
        my $max = 0;
        foreach (sort sortHash (keys (%hashName))) {
            if ($i == 0) { $max = $hashName{$_}; last }
        } # end-foreach
        
        sub sortHash {
            $hashName{$b} <=> $hashName{$a};
        }
    },
    'new' => sub {
        my ($max) = sort { $b <=> $a } values %hashName;
    },
});

print "\n\nComparison:\n\n";
cmpthese($r);


__END__

Timings:

Benchmark: timing 1000000 iterations of new, orig...
       new:  2 wallclock secs ( 2.44 usr +  0.00 sys =  2.44 CPU) @ 41
+0340.58/s
(n=1000000)
      orig:  8 wallclock secs ( 8.42 usr +  0.00 sys =  8.42 CPU) @ 11
+8736.64/s
(n=1000000)


Comparison:

         Rate orig  new
orig 118737/s   -- -71%
new  410341/s 246%   --
[download]

I hope that this helps. :-)

Update 31/05/2004: The Benchmark code I showed above seemed to be flawed. For some reason, using sub refs rather than eval strings seems to give better timings... the code was changed to reflect this.

$q=[split+qr,,,q,~swmi,.$,],+s.$.Em~w^,,.,s,.,$&&$$q[pos],eg,print