Also please remember that America isn't the be all and end all.
Mesa is an American company, I'm not going to talk about where something is distributed in Germany when the product manufacturer is an American company.
Orange- and Banana-pi's are still available with Allwinner and Rockchip SOC's with the integrated microcontrollers/PRU's. The original dev was more of a C dev than a Linux kernel for ARM and u-boot developer. Pine64 and Rock64 also have similar boards. When I get some time I'll pick a good kernel and build a proper u-boot for them. The rest of this project is mostly working. These were the devices that machinekit should have used, but oh well, they had some desire to use TI with a GPU that could not run a real time GUI.
Well it's hard to get people to see potential in something until it already exists in a useable form. I'd have already started writing the code if that was my thing but it's not. Coding for LinuxCNC is kind of obscure, I've worked with a friend on some LinuxCNC geared code but by the end I felt like I was pulling his teeth.
Personally I see the Teensy as a platform that would be good for a modular firmware. Mesa's HM2 is designed this way. The quality of the IO board is a moot point if the person writing the code didn't spend a lick of time on that specific PCB, they just wrote targeted software for the Teensy with the stepgen, encoders, as modules with some easily editable portion that resembles a "bitfile". I probably said all of that already but I do stand behind how useful that would be and the Teensy 4.x being the right platform for it.
You mention Machinekit, which I'll agree as I was never fond of BBB's. One lesser known thing MK did that could have been extremely useful was MKSOCFPGA. They ported HM2 to Altera for a DExx Nano. It basically treated the FPGA side as a Mesa 5i25 and all you had to do was modify a bitfile to give the pins any purpose they needed. The 2 problems that messed this up is the CPU in the Altera SOC was weak with no GPU at all, and this was not a thing that was supported in LCNC at all and MK was pretty much dead by the time I started looking at it.
I reached out to someone who definitely knows what they're doing but probably has no interest in this. I think if this got kicked off in a meaningful way it'd be pretty well adopted. It it's only seen as a cheapskate solution or a replacement for Mesa hardware in general it won't go anywhere because neither of those are good use cases.
