musing on StackInterpreter FreeRTOS VM target

 
Just sharing something thats been circling my head for a while
about a possible OpenSmalltalk based IoT platform.

Many micro-controllers support FreeRTOS
  https://www.freertos.org/RTOS_ports.html
including the ARMv7
  https://www.freertos.org/FreeRTOS_Features.html
and the ESP32 (based on Cadence's Tensilica Xtensa LX6)
  https://www.espressif.com/en/media_overview/news/espressifs-esp32-support-amazon-freertos

Of particular interest is Amazon's adoption of FreeRTOS
and its change of license from GPLv2 to MIT.
  https://aws.amazon.com/blogs/opensource/announcing-freertos-kernel-v10/

If a "thing" is not doing graphics, then the StackInterpreter may be
sufficient for many IoT applications. Having a FreeRTOS build target
for the StackInterpreter
might open the opportunity to have our Images running as the *only* application
on a wide variety of embedded systems.  A tight Edit-Compile-Run-Debug
cycle using the TelePharo remote tools might be very compelling versus
the usual Edit-Compile-Download-Boot-Run-BlindGuessDebug cycle.

I'd expect interest that starts with the hobby-maker community,
might later extend into the lucrative industrial control market
in competition with traditional PLCs.


Maybe getting the StackInterpreter running on a new platform
could make a good advanced-student project?


For starters, there seems a simulator here (I haven't used it)
https://www.freertos.org/FreeRTOS-simulator-for-Linux.html

which raises one question...
I understand the VM has a problem with some simulators
but is that more a problem for the JIT part of Cog than the StackInterpreter?

cheers -ben

 
Yes, this would be potentially very interesting. Small devices these days are so powerful they match fantasy engineering workstations of the early 90's. At least in principle it should be practical to run Squeak on an ARM SoC with 32Mb ram and anything better than 70MHz - hell, Eliot & I had the original ARM 1 at 4MHz and 4Mb ram running ST-80 in 1987, scoring near Dorado performance and it was *amazing*. A stackinterperet vm would be fine and easy to port (it's not like we lack experience at doing that) and you'd only need slightly bigger memory etc to make the Cog practical.

Craig has done lots of stuff around remote connecting and tiny images and building up what you need.

Find funding and go for it.

> On 12-07-2018, at 4:35 AM, Ben Coman <[hidden email]> wrote:
>
>
> Just sharing something thats been circling my head for a while
> about a possible OpenSmalltalk based IoT platform.




tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
Diagnostics are the programs that run when nothing else will.