Oh no! more damn linux stuff again!
The thing with MOS is that it runs on top of Quark kernel which user see nothing about when using MOS. So could be replaced with Linux kernel and user still not seeing anything about it. So for user nothing changes (except maybe with Linux kernel it would be a bit faster ... ;-p)
The advantage is that Linux kernel does not need to be ported/adapted to new hardware while with Quark it seems like every new hardware even if for user it looks like very similiar (Mac G4 version A <-> Mac G4 version B) needs work on the Quark kernel.
I wonder for example when a new port to new hardware was done - like Mac G5 - how much of work/time was needed to adapt/port Quark. And how much work/time was needed for the rest of MOS. Or in other words how much time/work would have been saved if porting/adapting the kernel was not needed.