This is something I've been wondering about. Porting from C to Rust one function at a time sounds nice, but presumably the end result would be a Rust program with an identical program structure to the original C, which as a result doesn't take very much advantage of Rust's features. Is the goal to port all the way to Rust and then refactor to actually make the code Rust-ish?
Yep! Even now it's not quite as bad as identical program structure. I mean, the high-level is fairly similar, but even now individual functions can look significantly different.
Emacs' C is already largely written in a high-level, macro-rich style to deal with Lisp objects (tagged pointers). So Rust already gives wins there w.r.t. more explicit types and From/Into impls over macros.
And over the last couple weeks there has been discussion about replacing xml.c with a native Rust backend for XML parsing.
