On Thu, Aug 22, 2013 at 3:37 AM, William ML Leslie < [email protected]> wrote:
> But we are talking about the platform that bitc bootstraps off, and how > that affects interface stability. You can start from scratch, or you can > start from an existing system, doing the latter likely makes it > multi-language by definition. > *You* may be talking about that, but I'm not. I'm looking at CLR as an alternative to LLVM. It's very plain that the CLR type system cannot describe the types that BitC wants to express at assembly boundaries. For those types that *can* be expressed, that may indeed permit cross-language access, but that capability isn't of any great interest to me. Just to note one big place where we have an issue, BitC doesn't currently have any form of inheritance. Objects in the CLR sense of the world can't readily be expressed in the BitC level type system. Unfortunately, I think that may be something about BitC that needs to change. I say "unfortunately" because it pushes us into some variant of F<, and that's probably going to reduce us to local inference. Which isn't the end of the world... shap
_______________________________________________ bitc-dev mailing list [email protected] http://www.coyotos.org/mailman/listinfo/bitc-dev
