From: gabriele renzi <surrender_it@...>
Date: 2005-12-07T02:57:34+09:00
Subject: Re: Shuffling an array, sort_by{rand}'s bias (was Re: need some Ruby

Uwe Schmitt ha scritto:
> || > || arr.sort_by{rand}
> || > that is not equivalent to my solution.
> || > the python equivalent to yours is
> || >   arr.sort(key = lambda x: random())
> 
> || > || > In Python it is quite easy by using list comprehensions...
> || > || What does that look like in Python? 
> || > 
> || > from random import random
> || > 
> || > def shuffle(a):
> || >     b = [ (random(), i) for i in a]
> || >     b.sort()
> || >     return [ x[1] for x in b ]
> || > 
> || > print shuffle(range(10))
> 
> || I would have thought those two functions do the same thing
> || with the exception that shuffle does not sort in place.
> || I'm probably missing something obvious. Could you explain
> || the difference between the two versions?
> 
> If you use random as sorting key, the result of a<=>b
> will differ each time you evaluate a<=>b. 
> In my case I attach a random number to each item,
> then I sort my list according to this number and remove
> this 'decoration' afterwards.
> 
> Pythonprogrammers call this pattern "decorate-sort-undecorate".

Are you sure of what you say? AFAIK
sort(key=lambda x: somefunc..))

will do decorate/sort/undecorate under the hood, behaving like ruby's 
#sort_by and like your version.
OTOH
sort(cmp=lambda x,y: something.. ) will behave like ruby's #sort.