sup talk - Dec 2007 - Util - Fixnum num_digits is wrong

Probably wasn''t causing any problems, but is wrong.

==================================================================---
lib/sup/util.rb     (revision 737)
+++ lib/sup/util.rb     (working copy)
@@ -299,7 +299,7 @@
 class Fixnum
   def num_digits base=10
     return 1 if self == 0
-    1 + (Math.log(self) / Math.log(10)).floor
+    1 + (Math.log(self) / Math.log(base)).floor
   end
   
   def to_character

-- 
Ian Taylor

Ian Taylor:
>  class Fixnum
>    def num_digits base=10
>      return 1 if self == 0
> -    1 + (Math.log(self) / Math.log(10)).floor
> +    1 + (Math.log(self) / Math.log(base)).floor
>    end
Math.log and Float#/ returning Floats means the above
approach is prone to the IEEE 754 representation inaccuracy.

Wouldn?t the below make more sense?

class Fixnum
  def num_digits base = 10
    to_s(base).size
  end
end

-- Shot
-- 
> Who wants to recommend a Usenet reader for little ol'' me?
I recommend Garrett Morris, particularly if you''re hard-of-hearing.
      -- jennifer friedoff and Joseph Nebus, rec.arts.tv.mst3k.misc
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 191 bytes
Desc: not available
Url :
http://rubyforge.org/pipermail/sup-talk/attachments/20071203/37b62dab/attachment.bin

Excerpts from Shot (Piotr Szotkowski)''s message of Mon Dec 03 02:52:10
-0800 2007:
> Math.log and Float#/ returning Floats means the above approach is
> prone to the IEEE 754 representation inaccuracy.
> 
> Wouldn?t the below make more sense?
> 
> class Fixnum
>   def num_digits base = 10
>     to_s(base).size
>   end
> end
Yes, but not as much as simply deleting the method. :)

-- 
William <wmorgan-sup at masanjin.net>