From: Vidar Hokstad <vidar.hokstad@...>
Date: 2008-03-29T23:55:07+09:00
Subject: Re: Why Ruby do not optimize code at all?

On Mar 28, 9:06 pm, Daniel Berger <djber...@gmail.com> wrote:
> Hm, couldn't you use method_added behind the scenes somehow? Perhaps
> use the opimized parse tree unless method_added is invoked? Or cache
> the parse tree?
>
> Or am I talking nonsense?
>

Not sure if method_added would be the right solution, but the
interpreter can certainly do checks. For the new VM based Ruby
implementations
for example, it'd be reasonable to do JIT based optimization of it -
it'd be relatively easy bookkeeping for the VM implementations to
check
for the replacement of a specific method of a specific class, and
throw away all JIT optimized paths relying on the behavior of that
method
and revert to the original byte code.

That could be done as a quite generic optimization, which pretty much
boils down to memoization of the results of frequently occurring
subexpressions that are known to be safe to memoize (no side effects /
stored state etc.)

Vidar