Wonder when is someone going to brave and fork the linux kernel and try to ffwd it with automatic programming.

why would you start there instead of creating something from scratch ?if you can port drivers just as easily meaning you don't especially give a shit about hardware you're running on in the first place, why even deal with linux? The battle tested LRU cache system?

	▲	literalAardvark 8 hours ago \| parent \| next [-]
		It's much easier to use something with all the edge cases already handled as a starting point.
	▲	convolvatron 8 hours ago \| parent \| prev [-]
		I've seen several workalike kernels in various stages of completion. at least one of them was able to run some pretty substantial applications (Postgres, nginx, that kind of thing), and that is still I guess around 250kloc. but it only really has drivers to support hypervisor devices. unfortunately as time goes by, the linux api surface gets larger and more convoluted. so there's going to be some coverage you're just never going to get. but in the abstract, definitely. linux is so bloated at this point that its not clear that it can ever be 'made safe'.