Well, I came to my senses and lost interest in booting and running from eMMC with NomadBSD.
What was I thinking? Just going from a USB3 stick to a weedy little on-board ssd makes a very nice difference. But going backwards to eMMC? Sure anything could be done - like running from cassette tape - but is it anything more than a technical exercise when all is said and done regarding desktop usability?
Awesome slacker excuse that!
At this point, the only way I could see eMMC being remotely usable for a desktop - other than a technical exercise - would be to load NomadBSD into ram, aka with a kernel option like “toram”.
And of course that means no ports - just package use unless you like burning up eMMC.
So, too many edge-case usability scenarios for me.
If I wanted to do this bad enough, I’d probably opt for a classic FreeBSD install instead, with a pared-down DE from the 80’s, maybe try to run from ram etc etc.
The Beaglebone is an ARM architecture, and to my knowledge there is no project support for that in NomadBSD (yet). However the same issue applies - in the end, will a typically low/slow resource truly be usable for you, or will it be just a technical exercise.
Again - maybe FreeBSD’s arm port might be a better solution in the long run. Here’s a guy who got it up and running in 2019:
But the same question arises - do you really want to run from eMMC, or the internal drive instead?