la_machine_2024-06-29.mp4
La Machine software is written in Erlang and runs on AtomVM virtual machine.
Please note that La Machine can only be flashed when it is not deep sleeping. You can action its button to wake it up. In case of necessity, reboot and reset buttons can eventually be used.
GitHub action builds an optimized image of La Machine including AtomVM virtual
machine with necessary codecs. It is artifact named la_machine.img of
Build workflow.
It can be flashed with ESP-IDF esptool on macOS as follows:
esptool.py --chip esp32c3 --port /dev/cu.usbmodem* write_flash 0 ~/Downloads/la_machine.img
The port may vary on your platform.
This image differs from standard AtomVM deployments as Erlang code is in a
single partition (boot.avm) with only La Machine code. This means that you
should not use rebar3 esp32_flash with default parameters (see below).
You need to flash AtomVM virtual machine with its libraries and
atomvm_esp_adf driver (the sound
codecs). You can build AtomVM from source or use the latest esp32c3 build of
atomvm_esp_adf's build action on main branch.
This build can be flashed with the following command on macOS:
esptool.py --chip esp32c3 --port /dev/cu.usbmodem* write_flash 0 ~/Downloads/atomvm-esp-adf-esp32c3-v5.2.2.img
To compile source code of La Machine, you need Erlang/OTP (a recent version
will do) and rebar3. With a single line, you can compile, pack and flash
La Machine code at default offset (0x210000):
rebar3 atomvm esp32_flash -p /dev/cu.usbmodem*
Alternatively, you can use the optimized build of AtomVM above and flash
La Machine. The partition map differ from standard AtomVM deployments and
you need to pack La Machine code with AtomVM libraries. You can use either the
full package atomvmlib.avm or
only estdlib.avm and eavmlib.avm.
rebar3 atomvm packbeam -p -e ~/Downloads/atomvmlib-v0.6.2.avm
rebar3 atomvm esp32_flash -p /dev/cu.usbmodem* -o 0x130000