Proposed Babel/SIDL Changes to Support RMI

RMI support in Babel has two main goals: transparency & flexibility. The additional RMI feature should be transparent to existing Babelized code, allowing painless upgrade. The RMI capability should also be flexible enough to support a variety of RMI transport implementations. The first goal's ideal would be for Babel users at some future date to simply upgrade to a RMI-enabled Babel release, regenerate files over their existing implementations, and find all their code is now able to be remotable without extensive modifications to their Impl files. The primary strategy for accomplishing the first goal is careful design and implementation of Babel generated code to minimize impact on user code. The second goal's ideal would be for Babel users to plug in robust WebServices-like modules when accessing Babel objects across a WAN, and utilize faster binary protocol for accessing Babel objects across a LAN, or even different nodes in a leadership-class supercomputer--without need to recompile their code. The primary strategy for accomplishing this second goal is to partner with appropriate parties to define an RMI API layer (in SIDL) such that various transport mechanisms can be ''plugged-in''.