Decide how to integrate with Rubinius' kernel/common

[judofyr]

I'd love to help out with integrating Rubinius' kernel/common, but I need some pointers.

• Should we use kernel/common?
• How do we handle the Rubinius-module? I vote for defining it under Topaz and then just set Rubinius = Topaz.
• How do we handle primitives? Primitives in rbx are kinda magic. Should we handle them at parse-time? Let the JIT inline them?
• How do handle the primitive names? E.g. array_concat, where should that method be defined? @classdef.primitive("array_concat", other="array")?
• Should the library files be shipped with Topaz, or bundled with the translated binary?

[timfel]

I'm sceptical about compiling it into the binary.

[timfel]

I would propose using a git-subtree, and removing the methods that do Rubinius.primitive. With a subtree we can track our changes to the kernel in this repository, and still merge upstream changes.

[alex]

I'm definitely going to be -1 on anything that needs primitives, also we should be moving away from embedding ruby in the binary, not adding more.