This is a proposal for a Linux init system based on MicroPython. Notice: included source code is an experimental, proof-of-concept working version to be available soon.

Proposed MicroPython-based init system, called mpy_init, is designed to run on bare metal and in containers.

1. Cross-platform – mpy_init inherits MicroPython's cross-platform nature, ensuring compatibility with all UNIX systems.

2. Target-based – mpy_init implements a current, target-based system initialization approach, similar to what is used in SystemD.

3. In between OpenRC and SystemD – mpy_init goes beyond the traditional scope of classic init systems like OpenRC and SysVinit. In addition to service initialization and control, it also manages:

   a. Time – ensures OS runs with a correct time.

   b. Network – ensures applications are capable of communication (necessary also for time synchronization).

   c. Periodic tasks – provides cron functionality.

   d. SSD trimming – Ensure partitions located on flash storage are mounted with a TRIM option, and/or periodic trims are enabled.

4. Parallel boot – boot services in parallel (if no dependency bound).

We believe that the above scope defines a good balance in what a modern init system should do, what not.

Micropython brings the following interesting features to initsystem:

API out-of-the-box — Micropython's VM is available since boot time throughout the entire OS run-time. Integration with init-system allows on bringing in API and web access that is handled by isolated VM. That simplifies device initialization and configuration over network, providing a layer of security.

Micropython lacks a built-in mechanism for process separation. Therefore, project elevates Unix kernel features to ensure proper privilabe and separation handling.

1. Performance - having MicroPython's VM already loaded makes scripts (compiled to bytecodes) to be run fast. Moreover, MicroPython is light-weight and optimized for resource-poor devices. Therefore, even in the case of running it on slow hardware, it should be a suitable solution.

2. Security through simplification - shell scripts can be replaced by Python code that is easier to read and audit.

3. Efficient development: debug and testing tools - Python provides robust development and testing tools.

4. More powerful shell - traditional shell (bash, ash) can be replaced with ipython or Xonsh - shells developed in Python. These projects keep compatibility with traditional shells (and theirs pros), while providing also Python features (even more Pros!).

• The MicroPython Project repository link.