You bring up a great point. Its a new day, and giving it another look I'd imagine that using rand() within a sort algorythm might be bad -- perhaps cause infinite looping because the comparsion between two items keeps changing. Apparently it doesn't though, but perhaps the default sort might with a larger set, or might change in the future. How's this?

I guess I should only do the rand once and memorize the result.

my @foo = qw( a b c d e f g h i j k l m n o p q r s t u v w y z );

my %comp;

my @feh = sort { if (! defined $comp{$a} )
                   { 
                   if (rand(1) >= .5)
                     {
                     $comp{$a} = 1 
                     }
                   else
                     {
                     $comp{$a} = 0;
                     }
                   }
                   $comp{$a};

              } @foo;

my $upper = int( scalar(@feh) * .75 );

print join (",", @feh[ 0 .. $upper]);

#C:\>perl randsort.pl  (output)
#b,a,d,c,f,e,h,g,i,j,k,l,m,n,r,q,p,o,s
[download]