Also fully fuzzy tested to not ever crash regardless of the bytecode given to it.
By small I meant like, at most only a few files in a single folder I can just drop into a project and be off running quickly and easily. Seeing cmake in that repo for example was a big sign that it's still way too big for what I want.
All the examples I've found seem to center around doing things interactively in the CLI. Do people write code like that? I write a lot of python and I maybe interact directly with a REPL for 5 minutes in a month.
There was no good tooling for managing a set of script files on the embedded device, and as soon as I had something mildly complex (on a ESP32, consuming from MQTT and writing to a serial device), I started hitting OOM errors. I rewrote a bunch of stuff to not use classes, which helped a bit, and then wound up giving up.
But to answer your questions directly, yes, I do use the REPL quite a lot to experiment and figure out how the system works.
I'm surprised that you had OOM errors, especially on an ESP32 where there's usually plenty of headroom. I haven't seen an OOM error - except when I've made a mistake - in years.
https://mitxela.com/projects/ddc-oled
DDC is how your OS reads the list of supported resolutions from your display and is present on VGA ports as well. It's just a low speed i2c bus and can be used for other things :)