Autor: o1bigtenor Data: Para: Devuan ML Assunto: [DNG] Fwd: Information request (maybe OT - - - dunno)
---------- Forwarded message ---------
From: o1bigtenor <o1bigtenor@???>
Date: Sat, May 13, 2023 at 7:16 AM
Subject: Re: [DNG] Information request (maybe OT - - - dunno)
To: Antony Stone <Antony.Stone@???>
On Fri, May 12, 2023 at 3:10 PM Antony Stone
<Antony.Stone@???> wrote: >
> On Friday 12 May 2023 at 20:43:12, o1bigtenor via Dng wrote:
>
> > I am in the beginning throes of an embedded system (SoC, and/or
> > microcontroller) design.
>
> I think that has become a vague definition over the years.
>
> Raspberry Pis were (and still are) a brilliant idea, but they got people to
> thinking you can run a full O/S on almost anything. Start working with an
> ESP32 and you'll notice the difference.
Am looking at something quite a bit 'lower' than a complete OS to start
although I'm not ruling out using something like the Pi if such horsepower is
warranted. Thinking of using the RP2040 (possibly in the RaspberryPi Pico
iteration) to manage sensors. >
> > Have been reading somewhat on the use of RTOS vs and OS (thinking
> > perhaps FreeRTOS and Devuan).
>
> My first question in a situation like this would always be "what exactly does
> this software need to run on?"
>
> If the hardware is a Linux-supported CPU with a sensible amount of RAM and
> access to a reasonable-sized storage device, then it's worth considering a
> full-blown O/S (and yes, for me that would always be Linux).
>
> If the hardware is more of a microcontroller (by which I mean a CPU which is
> not directly supported by standard Linux kernel builds, surprisingly little
> RAM (less than 1Gbyte), limited standard interfaces (SATA, USB, ethernet,
> etc.) or similar surprises) then an RTOS is probably the better way to go.
>
> If the hardware supports a "real" operating system, then installing and
> managing software is pretty simple.
>
> If the hardware is more basic than that, then you're just wasting limited
> resources by trying to run a full O/S on it, and you're better off starting
> from an RTOS or even just dedicated compiled code to do the job you need.
The RP2040 I'm considering not only uses a C kind of language but also
'micropython' which I believe is at least somewhat of a subset of Python.
Thinking of using that at least initially. If time shows that more
direct control
is needed. >
> Sometimes the choice starts from the hardware side "what can we get to run on
> this thing?" and sometimes it starts from the software side "we need to run
> this, what do we need to run it on?".
>
> It's not clear to me from your question whether you're being asked to find
> hardware which will run the necessary software, or being asked to work out
> what software can run on the given hardware.
>
> If neither is pre-determined, and you have a free rein to choose both, just go
> with what you know best, and then the project can be the most productive
> possible.
>
> In either case, please don't overlook the maintenance aspect - how do you deal
> with security patches, bug fixes, future changes in protocols (just look at how
> fast HTML changes, for example, and how badly out-of-date browsers deal with
> modern websites) in order to keep this thing functional for as long as
> possible.
There are even more challenging aspects.
One 'weighing system' is going on a mobile unit - - - - so it needs
easy ways of storing
data and of being useful even when its in a -35 or (hopefully very rarely) -40C
environment - - - how do I keep the system at a reasonable temperature
- - - how
do I keep it working even when its mobile.
>
> PS: I've always had a hard time working out what is actually OT for this list,
> and I like that.
> It was a long time mentor, now no longer with us for almost 4 years,
who suggested
first that I select Debian and then after systemd Devuan. He was by
his own term
'a dinosaur' yet worked in every aspect of programming and electronics
that I've
ever heard of (and a few I haven't - - - grin!). He gave these scary
long yet in depth
and highly understandable explanations for things. I'm finding that I miss him
almost more all the time - - - - he was a super resource. He was one
who did some
work in/on Devuan before it even got its name. His penchant for
accuracy is what
had me label the initial post as 'possibly OT' - - - am very thankful
I didn't just get
my ears burned.