Colin Watson wrote:
Since monolithic kernels are these days easily efficient enough for most if not all purposes, neither goal (of pulling a microkernel out of the Linux kernel or of writing a microkernel from scratch) attracts much attention.
But as I understand it, efficiency isn't the only reason for a microkernel.There's increased stability, as there are much fewer processes in the "core", and other processes are more strictly segmented from each other.
There's easier debugging/tweaking/modification, because of the more "modular" construction. Inserting a modification to the kernel doesn't require a reboot, and can't hose the system if the modification crashes.
Provides freedom to the user. If the user wants to use the xyz filesystem, he doesn't have to beg the system administrator to enable it in the kernel; the user can just plug it in himself.
So even though microkernels may not attract much attention currently, I think that's more due to a lack of awareness of what a microkernel offers and to the lack of a system (HURD) that's beyond the alpha stage. If the HURD ever gets usable, it might start attracting attention.
Of course, I could be an idiot. -- Kent