From: Charles O Nutter <headius@...>
Date: 2006-09-29T08:51:29+09:00
Subject: Re: Obfuscation (was Re: Ruby with Qt or GTK ?)

On 9/28/06, Ken Bloom <kbloom@gmail.com> wrote:
> Ruby can't even be really effectively obfuscated (at least not as
> effectively as Java), because it depends on things being named the same
> for duck typing to work. In Java, you can rename any method in the
> inheritance chain that isn't defined in the parent class, you also get
> overloading to help with the obfuscation so you start by renaming
> functions to as many overloads of "a" as possible, then to as many
> overloads of "b" as possible and so forth. In Ruby, you don't have the
> naming heuristic, nor do you have the ability to create many overloads of
> the same function.

Java includes all naming in the compiled bytecodes as well, because
it's an integral part of the method signatures. Java code can be
decompiled to almost exactly what was fed in, with few exceptions.

Under JRuby, Ruby code will probably be obfuscated even more than Java
code. I've been working on the compiler for JRuby, and I guarantee the
resulting bytecode won't be easily reversible. You might be able to
reverse it to some pretty heinous Java code, but reversing back to the
original Ruby source will be practically impossible. Even if the
method names are still present, everything else will be Java
bytecodes, many of them specific manipulations of JRuby's runtime.
Ruby's AST itself can be reversed, but not perfectly, and if passed
through a compiler like the one in JRuby, it will be mostly
unrecognizable.

-- 
Contribute to RubySpec! @ www.headius.com/rubyspec
Charles Oliver Nutter @ headius.blogspot.com
Ruby User @ ruby.mn