Cool idea. stefp++.
Curious if sort could get into a 'never finish' scenario if the right values came up from rand though?
Thought you might like this version of the fisher yates shuffle.

sub fy_shuf{my($l,$c,$r)=@_;map{$r=int(rand($_+1));
@$l[$_,$r]=@$l[$r,$_]}reverse 0..$#$l for 0..$c||0};

my @a=(1..52);
fy_shuf(\@a,5);
[download]

UPDATE
Having had a look at japhy home node I borrowed an idea from him and reduced it more. (Oh and I decided that shuffling mutiple times was unecessary.)

sub fy_shuf{my($l,$r)=@_;map{$r=int(rand(@$l-$_+1));@$l[-$_,$r]=@$l[$r
+,-$_]}0..$#$l};
[download]

Yves
--
You are not ready to use symrefs unless you already know why they are bad. -- tadmc (CLPM)