From: Evan Phoenix <evan@...>
Date: 2007-07-13T16:44:37+09:00
Subject: Re: Proposal: runtime-modifying Kernel methods should be keywords

I'm in complete agreement with Charles on this. In fact, the rubinius  
compiler already short circuits a few of these, giving them keyword  
behavior (private and public with no argument, block_given?).

I believe that the primary criteria is methods that modify the  
runtime behavior of a methods scoping. All of the methods given so  
far have this in common.

To address Marcel's wish for essentially more first class data  
structures in Ruby, so that things weren't happening 'behind the  
curtain', in a dark place you can't access. As a developer of the  
ruby implementation with the most first class data structures, it's  
still very valuable to have these as keywords. Their presence in a  
method governs how the method can run the best, so easily detecting  
them is paramount.

In other words, this isn't a question of how can we continue to hide  
those details, but rather how can me make those details easier to  
implement quickly and safely.

  - Evan


On Jul 12, 2007, at 11:33 PM, Yukihiro Matsumoto wrote:

> Hi,
>
> In message "Re: Proposal: runtime-modifying Kernel methods should  
> be keywords"
>     on Fri, 13 Jul 2007 15:02:06 +0900, Charles Oliver Nutter  
> <charles.nutter@sun.com> writes:
>
> |I propose that a number of Kernel methods should be keywords.
>
> I'd like to hear opinions from others, especially from other core
> implementers, Koichi, Evan and a few more others, before making final
> decision.
>
> 							matz.
>