From: Dave Thomas <dave@...>
Date: 2004-03-06T12:17:58+09:00
Subject: Possible strangeness in Numeric

To my surprise, Numeric defaults to implementing to_int by calling 
to_i. This means that, for example,

    (1.2).to_int  => 1

I don't think this is correct, as a Float is not a representation of an 
Integer, and so can't really stand in where an integer would. But 
having Float respond to to_int by rounding itself down implies that 
Floats are usable as Integers.

     obj = IO.new(3.14159)

That just doesn't seem right.

Clearly Float needs to support to_i, but to have all the Numeric 
objects (including Float and Rational) support to_int is (I believe) a 
violation of the intention of the to_xxx methods.

I suggest that the to_int implementation should be moved from Numeric 
to Integer.

Cheers

Dave