The original document is written in Chinese so far, and I'll release an English version with further development. But the major recognizable difference is the RAII-wrapper for many synchronous primitives like mutex/semaphore/message queue/spin-lock, also I've been trying to add async programming mechanism that is similar to those in Rust by taking advantage of C++20 coroutines, including an trivial implementation of future<T> that needs no dynamic allocation(which is crucial in embedded system context).
By now, only Cortex-M serires are tested, but the transportation to RISC-V(ESP32C-x series for a start) platform has been added to future todo-list, since I've split the layer between hardware abstraction and software kernel into `vendor-hal` and `kernel`, along with the user application layer `user`.