Amazing write up! Everyone at work is itching to try Rust, but I think what’s killing adoption is that it’s not very clear how to gradually transition a code base. We have a few million lines of C++, some of it written 25 years ago. A full rewrite is just out of the question, at best we could use it for new sections. This is super common in the c++ world, so it’s a pity that porting wasn’t a first class concern in rust considering C++ devs are the target audience. It sounds like it was a challenge even at 57k LOC. Congrats to the fish team though, great accomplishment!
If you codebase isn't somewhat modern C++(C++11) I would start there before concidering a port to rust. It will be a significantly easier upgrade in safety even if not going all the way to rusts level of safety.
Generally code that has been running for years is unlikely to have too many bugs since they have been shaken out, "rewrite it in Rust" as a fad just ignores the decades of work already put into the codebase and for large codebases likely eont succeed.
As you mentioned, write new modules with rust. That means likely needing to export a C API for your libraries but there's a good chance you were already doing that. There was also a rust crate that tried to automate most of the c++ rust interop for you but not sure about how good it is in reality.
