🚀 go-pugleaf

On Saturday 05 August 2006 19:18, Vladimirr Saigon stood up and spoke
the following words to the masses in /comp.os.linux.misc...:/

> Does anybody know why Linus choose to make linux monolithic instead of
> microkernel? I have been told, that there was some dispute about that
> decision.

In short: monolithic kernels are faster and less prone to errors caused
by the increased complexity or kernelspace/userspace signalling. 
Microkernels require a lot more semaphores to help userspace and
kernelspace communicate, and this typically impedes performance.

Aside from that, the reference provided by the poster known as /yttrx/
will give you a more in-depth view on the debate between Linus Torvalds
and Andrew Tannenbaum - the creator of Minix.  

Minix is indeed microkernel-based - there are some who would dispute
that - while Linux is a monolithic kernel.

Perhaps also noteworthy is the fact that native GNU as intended by
Richard Stallman also makes use of a microkernel design, but that this
design was so complex that it was the last component of the operating
system to get developed.  At present, the Hurd is still not considered
production-ready.

As Richard Stallman's intentions were not only technical but also - and
perhaps foremost - politically, he doesn't really mind that the Hurd is
catching up late.  After all, his political goal - i.e. the
availability of an entirely Free operating system (in the sense of
Freedom) - has been accomplished by coupling the Linux kernel to the
GNU operating system userspace.

It's an obviously very efficient and very stable combo, and one should
never change a winning team. ;-)

-- 
With kind regards,

*Aragorn*
(Registered GNU/Linux user #223157)