"I'd be happy if they did break compatibility and provide a sand-box for old apps. While they're doing that, they might as well break other stuff to allow SMP, and to lock off user programs from accessing OS structures. "
I agree with you. Of course this brings the conversaton back to the forking argument. The point of not breaking compatibility is so old software could continue to run. Right now, we have to rely on AInc to run most old software because they own the Kickstarts. My primary interst in AROS is that I keep hoping that it will eventually get reported to the 68k, and we could finally get rid of AInc.
I guess the way I see it is that for any old software, I want to be able to run it on my MiniMig, one of it's successors, and maybe UAE. For new software, it doesn't matter if we break compatibility with the old API.
To answer the original questions:
1. Can it be done?
Yes.
2. Can it be done elegantly?
Sure, if Aros was ported to 68k so that we don't have to go to AInc to get kickstarts, and UAE were integrated so that it worked out of the box using that port it be plenty elegant.
3. Should it be done?
That is the question. I would say it would make sense, but that the 68k port is far more important than it is given credit for.