<rant> I think you're right that there's no viable alternative to C for IOT right now. However, I also think we need to collectively move past C in this space. The compile, run, debug cycle in embedded electronics is just too slow. I don't really mind the language, just the painfully slow development (and lack of package management). uPython and similar are a move in the right direction, but I think we need a minimalist crossplatform bytecode runtime to become standard on every popular embedded platform. This runtime needs frictionless c FFI (with the ability to seemlessly drop into c at any point, though it may require a recompile), and to be designed with embedded in mind. Actors and RPC should be first class citizens, as should remote inspection of systems over the wire, hotswapping running code, and the security needed to do those things safely (easy to get started with automatic self signed encryption). I think that every attempt at doing this so far is focusing on creating hobbyist boards for people who don't want to learn c instead of creating this ideal runtime with seemless c ffi (allowing easy adoption on new platforms), or is not lightweight enough to run on $4 chips like the ESP32. I think that most likely this platform will need to be written in c, and provide simple hooks for tying the networking libraries for individual boards together.

I was looking into building this for something my company needs, but then a Taiwanese company approached me about a low cost industrial modular linux device with hotswappable i/o they're developing. At this point I just shrugged and figured that seemed a lot easier than building a better wheel.

The thing is, I can use a $400 modular linux device because my market can bear it, but I was looking at developing the other set of features for inexpensive micros. Virtually every market can bear the cost of a $4 chip. </rant>

I apologize for the rant, it just physically pains me that this doesn't exist, and that embedded development is so primitive.