Re: Is there any plan to make luajit support bytecode?

[Peter Cawley <lua@...>]

On Fri, Apr 15, 2011 at 11:05 AM, Adam Strzelecki <ono@java.pl> wrote:
>> It also appears that the speed of loading precompiled code is a factor
>> in games;
>
> Agreed. It would be worth to check what's the parsing cost time in some real project scenarios with lots of scripts.
>
>> bytecode itself is not good obfuscation for a determined
>> person with a dissembler.  If compile speed is not a factor, then some
>> traditional code obfuscation would probably be as good as bytecode.
>
> There's no perfect protection. It is always about the ratio between time/costs of breaking the protection and the potential gain of doing so. In case of bytecode compilation these costs are higher, as breaking it requires much more skilled and determined person to disassemble than when breaking source code obfuscation, which can be "de-obfuscated" and reformatted pretty easily. Finally binary bytecode files discourage most of potential "hackers" from plumbing your code.

In the case of Lua source, the output of a good automated disassembler
ought to be similar to the output of a good automated obfuscator. To
add another layer of protection, you could encrypt the obfuscated
source, and then decrypt it a chunk at a time in the reader function
passed to lua_load.