In terms of FFI, long-term I do think we want better integration for objects with “foreign lifetimes”: Python objects managed via Python reference counting, JavaScript and DOM objects in a browser, objects managed by a foreign garbage collector, EFI firmware objects, and other “external” objects. I’d love to have a better solution for those than just “write a wrapper using a raw pointer”, when a large family of otherwise unrelated objects all use the same “foreign lifetime” mechanism.
However, I don’t think that should go in the 2017 roadmap. For 2017, I’d like to see exceptional support for C FFI: better library integration, better build system integration, improved cdylib support, sonames, and anything else needed to transparently put Rust anywhere C can go.