From: Rick DeNatale <rick.denatale@...>
Date: 2006-08-10T05:29:28+09:00
Subject: Re: nextPowerOf2(n)

On 8/7/06, Simon Kr�ger <SimonKroeger@gmx.de> wrote:
> Rick DeNatale wrote:
> > On 8/7/06, Simon Kr�ger <SimonKroeger@gmx.de> wrote:
> >> Rick DeNatale wrote:
> >
> >> > My original intention was to benchmark the various suggestions, but
> >> > since you can get the wrong answers with arbitrarily fast code, I
> >> > don't think that it matters much.
> >>
> >> Hmm, what about:
> >>
> >> def next_power_of_2(n)
> >>   throw 'eeek' if n < 0
> >>   return 1 if n < 2
> >>   1 << (n-1).to_s(2).size
> >> end
> >>
> >> ?
> >
> > Ahh, but the following is both quicker and more economical of source code
> >
> >  def next_power_of_2(n)
> >  end
> >
> > If you don't care about a correct answer, then nil is as good as any. <G>
>
> would you dare to explain?
>
> def next_power_of_2(n)
>   throw 'eeek' if n < 0
>   return 1 if n < 2
>   1 << (n-1).to_s(2).size
> end
>
> puts next_power_of_2(0)  #=> 1
> puts next_power_of_2(1)  #=> 1
> puts next_power_of_2(2)  #=> 2
> puts next_power_of_2(3)  #=> 4
> puts next_power_of_2(4)  #=> 4
> puts next_power_of_2(5)  #=> 8
> puts next_power_of_2(255)  #=> 256
> puts next_power_of_2(256)  #=> 256
> puts next_power_of_2(257)  #=> 512
> puts next_power_of_2(536870911) #=> 536870912
> puts next_power_of_2(536870912) #=> 536870912
> puts next_power_of_2(536870913) #=> 1073741824
>
> seems to be correct for a lot more cases than nil(?)

You had to go a bit further back in the thread.  All of the versions
which relied on float math functions fell apart for differing
increasing values of n as the floating point math starts to diverge
from 'pure' math.

I never said that it wasn't possible to write a method which produced
accurate answers, only that those float based ones don't and that
there was no sense in benchmarking implementations which produce bogus
results against those which do.

-- 
Rick DeNatale