In the early stages of any programming language, one of the biggest problems to user adoption is the lack of libraries. With that in mind, any platform that offers “free” and efficient interoperability for the languages that can run on it (simply by being compilable to LLVM bitcode) deserves a close look because it would make so many specialized packages instantly available for use. From what I understand, the kind of FFI available to languages hosted on Graal is significantly easier to achieve and is likely more efficient than other methods.
As far as empowering/collaborating with Luna in the sense of participating in the running graph, I would think external library functions typically act as either source or sinks for data in the flow rather than transforming it, but maybe it could be relatively “easy” (due to the interoperability mentioned above) to instrument an LLVM bitcode blob that’s the result of compiling some arbitrary other language with the proper input and output ports so that it could participate anywhere in the graph.
It might be an interesting thing for someone really familiar with Haskell to try and compile targeting the LLVM. On the GHC page I see that the “GHC compiles Haskell code either directly to native code or using LLVM as a back-end”. I have no idea how much work would be involved in trying to get everything needed into bitcode and maybe it’s a really big task. Maybe not though, and if reasonable to do, the short term benefits could be very significant. We’ve already seen people asking for R, Python interop. If not done through a unifying platform, all of these would involve considerably separate efforts. On Graal, there would be interop with several languages from the start, though it’s true some are in early stages of development.