From: Daniel Schierbeck <daniel.schierbeck@...>
Date: 2006-06-20T04:10:57+09:00
Subject: Re: Is it more convenient to have a random_each in ruby stdlib?

Matthew Smillie wrote:
>> Since #shuffle! would alter the receiver, it shouldn't be defined in 
>> Enumerable. Other ordered collections could implement their own versions.
> I agree with you there, but it's not quite what I meant.  More clearly, 
> Enumerable#entries returns an array with no guarantees about the 
> ordering.  Enumerable#sort and #shuffle return similar arrays, but 
> include guarantees about the ordering.  Enumerable#shuffle uses 
> Array#shuffle! internally to produce the final array, which just doesn't 
> quite click for me.  It's just an aesthetic disagreement, and I can't 
> quite put my finger on anything concrete.

I can see what you mean -- you'd rather have #shuffle! depend on 
#shuffle than the other way round? I.e.

   class Array
     def shuffle!
       replace(shuffle)
     end
   end

I guess that could be possible.

   module Enumerable
     def shuffle
       ary = entries
       ary.each_index do |i|
         r = rand(ary.size - 1) + i
         ary[i], ary[r] = ary[r], ary[i]
       end
     end
   end

Of course, it would not be much different than `entries.shuffle!'


Cheers,
Daniel