> mrustc is not the solution [...] as it’s only capable of building rustc 1.39 This seems like an odd point. Instead of updating mrustc to support a couple of

> mrustc is not the solution [...] as it’s only capable of building rustc 1.39 This seems like an odd point. Instead of updating mrustc to support a couple of more features every once and a while we should write a whole new frontend? It seems like they are basically solving the same problem, except mrustc has chosen a subset of it (it assumes correct code). Of course GCC-rs also has a subset in a way as most of the codegen is handled by the GCC core. To be clear I'm not saying that mrustc is a better approach to bootstrapping, but for this specific point is seems that the two approaches are basically equivalent.