From: matz@... (Yukihiro Matsumoto)
Date: 2003-07-19T20:48:01+09:00
Subject: Re: chaining comparisons

Hi,

In message "Re: chaining comparisons"
    on 03/07/19, news_chr <swap@gmx.net> writes:

|class Fixnum
|  alias __lt <
|  alias __le <=
|  alias __gt >
|  alias __ge >=
|  def <(r)
|    __lt(r) ? r : nil
|  end
|  def <(r)
|    __lt(r) ? r : nil
|  end
|  def <=(r)
|    __le(r) ? r : nil
|  end
|  def >(r)
|    __gt(r) ? r : nil
|  end
|  def >=(r)
|    __ge(r) ? r : nil
|  end
|end
|
|
|class NilClass
|  def <(r)  self end
|  def <=(r) self end
|  def >(r)  self end
|  def >=(r) self end
|end
|
|
|p ((4 < 5) < 6) < 7  # 7
|
|p ((4 < 5) < 5) < 7  # nil
|
|p ((4 < 5) >= 5) < 7 # 7
|----
|
|Personally I wouldn't mind such a semantic but it is
|probably too wired for general consumption ...

They used to work like this long time ago.  But I felt the semantic
was too weird.

							matz.