I am planning to write a new L4-based multi-server OS called UX/RT
(initially I will use Fiasco.OC as the kernel since it has x86_64
support and supports kernel capabilities, but may eventually support
seL4 as well). Unlike the other current L4 OS environments, it will be a
multi-server Unix-like OS resembling QNX and Plan 9. I am intending it
to have good support for both desktop and embedded systems (servers are
not a huge priority; the microkernel architecture might limit the
maximum throughput for heavily I/O-bound workloads). I am dissatisfied
with all the current mainstream OSes and want to take on the challenge
of making something better. I am planning to reuse as much existing
third-party code as possible, especially as I will probably be working
on it alone at first (of course if anybody here wants to contribute they
can). However, due to the design, the only part of L4Re that I can use
as-is is sigma0 (which may require minor changes to fit into the
XIP-based boot process I am going to use). I will be implementing my own
root task which may use a few parts of moe, although I haven't looked
too closely at the source for moe so I'm not sure if there's anything
there that I can use. Other than sigma0, the low-level kernel headers,
and possibly some parts of moe, I'm not sure if there is much else from
L4Re (or Genode) that would be useful in the context of a pure Unix-like
multi-server OS. Most of the code I reuse will come from BSD and GNU
(which gets me quite far; the root task will probably be the biggest
part of the base OS that will be primarily original code). I already
have a very good idea of the design I'm going to use; I'll post more
about it later.
