From: Alan Chen <aero6dof@...>
Date: 2004-07-17T05:42:13+09:00
Subject: Re: Dynamically replacing methods for efficiency


Hal Fulton wrote:
> On the one hand, I appreciate Ruby because I *can* do this
> sort of thing. On the other hand, if the logic for #action
> changes, how many pieces of code will I find myself modifying?

Exactly, you *never* want to code yourself into a corner where one
change in logic needs to be refelected in parallel branches -
especially when
your involving diagnostic code. DRY!

A different way to approach your problem is to have a smarter tracer
object which performs a noop when its inactive.

def action
@tracer.puts "Entering action"
do_this
@tracer.puts "Returned from do this"
do_that
@tracer.puts "Returned from do that"
end

Your trace setter would activate the tracer which would swap between
a noop puts method and a working puts method. Now, I've aproached
things this way in C++ where a noop is very cheap, I'm not sure of the
execution cost of a method lookup vs an if statement in ruby. Either
way, I think it's still conceptually cleaner to use a smart tracer,
logger object.