Roc on microcontroller?

[elrom]

Has anyone tried running roc on a microcontroller on bare metal? Authors - any idea how hard it would to get that running?

[baranovmv]

Hi elrom, Thank you for your interest. At the beginning we planned to keep the codebase self-sufficient to be able to run on bare-metal embedded or some kind of RTOS. However, in the course of development roc adopted an asynchronous approach for multithreading programming, and uses libuv internally. Which is the main blocker for compiling for bare metal embedded platforms. In my personal opinion, there are no some isolated spots in the codebase which potentially could be somehow tweaked to be able to run with say lwip instead of posix compliant standard library. If you are really keen on this task, I suppose you'd better take the processing pipeline code of the roc and redo the environment. If you wish, we could discuss this approach in more detail here

…

On Mon, 29 Aug 2022 at 01:23, elrom ***@***.***> wrote: Has anyone tried running roc on a microcontroller on bare metal? Authors - any idea how hard it would to get that running? — Reply to this email directly, view it on GitHub <#478>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAO4V53KP7VY2O2QTCZTVVLV3PKD7ANCNFSM573WZMFQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

-- Best regards, Mikhail Baranov, +7(926)559-44-01 ***@***.***

[gavv]

We still have pipeline and I/O layers very well isolated (they communicated via tasks and queues), as well as platform-specific code is isolated in target directories.

I think running on bare metal fits current deign well (i.e. the whole pipeline may be reused), but of course this will require quite a lot of work (reimplementing network layer for example).

[gavv]

See also #496