Björn Quentin

Interesting idea - I didn't think of that (while the constructors do it the exact same way 🤦‍♂️ ) .... since this is an optional addition, I'm probably fine with...

Yes, that is the kind of comment I was looking for 👍 >I'm not an expert on this sort of thing, but isn't that what PhantomData could be used for?...

> Do you happen to have a good resource on implementing type state? I took the one from the embedded-rust book as template. There are quite some articles on it...

Sure - I will add the things I identified - might not be an exhaustive list however

What should get changed in a way that runtime-errors are turned into compile-time errors (or make it error out at all) __EDIT:__ task list moved to first post so that...

There no specific reason other than "nobody did it, yet"

note: the upcoming LEDC implementation is currently also using floats

No problem with that - we can replace the floats in one go later

I currently don't have a solution for your problem but I agree `xtensa-lx::Mutex` is inconvenient to use In general I guess we should avoid to use `xtensa-lx` / `riscv` types...

> In fact, it seems to be a re-export of mutex_trait::Mutex as seen [here](https://github.com/esp-rs/xtensa-lx/blob/3e2120aaea0e83dd6e949cdaa24d16edd07632c8/src/mutex.rs#L5), so it isn't specific to the architecture. Yes, it's that trait ... which is not implemented...