From: Marko Schulz <in6x059@...>
Date: 2002-03-20T23:12:06+09:00
Subject: Re: Float comparison

On Wed, Mar 20, 2002 at 06:49:49PM +0900, rubydev.deweerdt wrote:
> Hi,
> I'd need to compare floats with a choosen precision, a sort of :
> 
> 3.1415 == (2) 3.1416   #The (n) being the precision
> Would be true
> 3.1415 == (4) 3.1416
> Would be false
> 
> Is there a way in Ruby to accomplish this ?

You could redefine Float#eql?

  class Float
    def eql?(other, delta = 0)
      (self - other).abs <= delta
    end
  end
  
  p 3.1415.eql?(3.1416, 0.001)   #-> true
  
  p 3.1415.eql?(3.1416, 0.00001) #-> false
  
  p 3.1415.eql?(3.1415) #-> true
  

To prevent some errors concerning rounding it may be better to handle
the case of delta==0 special:

  class Float
    def eql?(other, delta = 0)
      if delta == 0
        self == other
      else
        (self - other).abs <= delta
      end
    end
  end
  
Then again I am a little bit afraid of a recursion trap when defining
.eql? with ==, since I do not know Rubys internals.

-- 
marko schulz