From: matz@... (Yukihiro Matsumoto)
Date: 2001-08-24T11:40:36+09:00
Subject: [ruby-talk:20236] Re: ++ Operator

Hi,

In message "[ruby-talk:20234] Re: ++ Operator"
    on 01/08/24, Dave Thomas <Dave@PragmaticProgrammer.com> writes:

|OK, but... (this is where I make a fool of myself).

You sacrifice yourself for the sake of our enlightnment, don't you?

|Is the fundamental property of Fixnum that its objects are immutable,
|or that its objects do not have copy semantics? If the former, then
|obviously ++ just isn't viable. If the latter, then it's possible to
|imagine implementations that would allow it.
|
|I know we're not going to see ++ in Ruby in my lifetime. However, I'm
|wondering if this is a necessary restriction on the language, or
|simply on implementation convenience.

To implement "++" behavior, we have two choice:

  (a) make "++" a special form of assignment.  "a++" would be syntax
      sugar for either "a+=1" or "a=a.succ".

  (b) make numbers mutable, then prepare "++" method to alter the
      number value.

You were talking about (b) above, right?  I think it's wrong way to
go.  It is confusing. I believe almost everyone assume numbers to be
immutable consciously or unconsciously.  Also it would be performance
penalty.

If we really want "++", (a) is the way to go, but I don't think it's
worthy to add new assignment syntax.  Mostly because we have less
chance to use it in Ruby.  Thanks to iterators.

							matz.