From: matz@... (Yukihiro Matsumoto)
Date: 2004-03-08T13:20:25+09:00
Subject: Re: Possible strangeness in Numeric

Hi,

In message "Re: Possible strangeness in Numeric"
    on 04/03/07, Mathieu Bouchard <matju@sympatico.ca> writes:

|Have you tried something like this?:
|
|class Numeric
|  def to_int
|    x=self.to_i
|    raise "not a representation of an integer" if x!=self
|    x
|  end
|end
|
|class Float; remove_method :to_int; end
|class Integer; remove_method :to_int; end
|
|So (2.5).to_int raises, but (2.0).to_int returns 2, and
|Rational(4,3).to_int raises, but Rational(6,3).to_int returns 2.

Very interesting idea.  I'm not sure whether this is what people (who
wants float values to integer arguments) want.  But still very
interesting.

|Complex(2.0,0.0).to_int, however, doesn't work, because Complex doesn't
|have a #to_f nor a #to_i at all. I am wondering what a good default #to_f
|for Complex would be. It wouldn't be modulus because then
|Complex(-2.0,0).to_f == +2.0, which would be really bad. Taking the real
|part sounds more reasonable. I don't really have any other ideas.

We can just raise TypeError Exception when number does not have to_i.

							matz.