Hmm. By
not providing a POSIX lemonade stand, would-be users are required to basically figure everything out for themselves; I wonder what the net impact of that is.
On the one hand POSIX et al is effectively impossible to deploy in a secure way, so there is a reasonable argument for going back to the drawing board; but on the other hand there isn't really a well-defined go-to alternative How To Computer model that is friendly to provable security, so everyone has gets to reinvent that wheel every time
Considering the contemporary status quo in terms of independently-implemented OS projects and platforms (eg, my ever-so-slightly-wobbly VxWorks-based TP-LINK consumer ADSL modem), I do wonder how good seL4 implementations end up working out in practice - the kernel might be rock solid, but what about all the bits on top of it, some of which presumably communicate with the outside world, consume various protocols, need to control the hardware in various ways (which includes relying upon reading the hardware state/status), etc?
