From: "trans.  (T. Onoma)" <transami@...>
Date: 2004-10-29T05:09:25+09:00
Subject: Re: Ruby Vs. ... might find comparison of interest.

On Thursday 28 October 2004 03:49 pm, James Britt wrote:
| > So its easy to say he doesn't know diddly. But we as a community still
| > need to address such things --and from both sides. Where they are
| > accurate we should improve on our end, and where they are inaccurate we
| > need to get them to improve on their end. On that not I will email him.
| > Is ther anything else I should add besides the instance var thing?
|
| I already sent him a note.  We had short exchange and he is going to
| update his Ruby-DyBase API.

You da Man!

| > Speaking of which what advantage does
| >
| >   obj.instance_var_set("@x", 10)
| >   obj.instance_variable_get("@x")
| >
| > have over,
| >
| >   obj.@x = 10
| >   obj.@x
| >
| > Surely the later can be parsed.
|
| Perhaps.  I'm at odds over the ability to simply trod over the
| object->message-to-access-an-attribute approach, where one must
| explicitly expose object attributes via messages if other code is to
| access them.
|
| It can be easy to abuse, but then, so what?
|
| The uglier syntax may at least serve as a red flag that you are doing
| something of dubious OO purity.  If it matters.

I pretty much lean toward the "so what" here. It's a bit silly to, on the one 
hand, support a feature and yet, on the other, make it syntatically awkward 
b/c it goes against some "formal OOP grain" --a valid concern, but these are 
indispensable for meta-coding. So, I think the more appropriate way to handle 
it is by having to require the ability first.

  require 'iv_accessors'

my 2%,
T.