Is Embedded Systems development ripe for a jump in productivity? 30' with Nerves Project

In a chat on the refactoring community somebody mentioned Nerves Project as an alternative to the reigning programming language (C) in embedded systems.

I was at the airport with 30' to kill and decided to check it out and see how I could go in just half an hour.
Turns out I managed to fully compile the hello_nerves project and stopped short of flashing the image to a rpi3 which I did not have with me.

I find this pretty impressive, considering it's a fully-fleshed Linux-based system for half hour of work on a flaky internet connection.

Installation is pretty straightforward and I love that they support asdf (which has been my go-to tool for managing environments and tools for the past 6 years). Once I got Erlang and Elixir installed I just had to create the project:

mix nerves.new hello_nerves
mix archive.install hex nerves_bootstrap
cd hello_nerves
export MIX_TARGET=rpi3

realizing that I needed an extra command or two to fetch the dependencies took a couple of wrong turns, but I was bale to figure it out eventually:

mix deps.get

and then I was able to finally compile and build the firmware image:

mix firmware
mix firmware.image

Not having an rpi3, I thought that maybe I could create a VM that emulates it. For some reason that did not work out. The VM booted but did not show anything on the console, the qemu processes qemu-system-aarch64 were just running at 100% which suggests a lockup.

Conclusion

It's simply mindblowing how much progress I could make in half hour, and I'm sure the power of Elixir and reliability of Erlang bode well for the success of this project.

Nerves sports a firmware update system too, so you really get a lot of value out of the box.

Where I'm a bit concerned is in the available support: Erlang and Elixir are a relatively niche community which does not even register on the most recent Github stats.

To test this theory I went and looked at BLE support (which I would expect to be a relatively important subsystem to have) and, unfortunately, there seem to be a only half-supported library which claims supports for a couple of chips.

Maybe fine for a POC or a hobby project but I would need a well-argued narrative to convince me to make a professional product on top of Nerves.

On the other hand this shows pretty clearly how Embedded Systems market is ripe for change. While nobody might have never been fired for using C, those willing to get ahead will have to look elsewhere.