[linux-elitists] Freeing parts of Unix?

Aaron Sherman ajs@ajs.com
Thu Jun 14 06:34:41 PDT 2001


On Wed, Jun 13, 2001 at 11:24:53PM +0200, Eugene.Leitl@lrz.uni-muenchen.de wrote:
> "Mr.Bad" wrote:
> 
> > Even with the kernel... does anyone really think there's some secret
> > surprise goodness left in the Unix kernel? Except for some funny
> 
> Yes: there's too much secret goodness left in the kernel. So let's
> chuck it all out -- all of it -- and go concurrent message passing
> nanokernel. Few kBytes, written in 100% hand-coded pure machine 
> language, as God intended kernels to be.

Go for it. No one has ever succeded in creating such a system that is
not painfully inefficient, but I welcome you to try.

I see the whole micro/nano kernel vs. monolithic thing to be a red
herring. Take, for example streams in System V. This is neither the
monolithic nor the micro approach, but a hybrid which allows building
device driver-like, runtime-loadable modules (like, e.g. Linux) in a
fixed framework of OS/module interaction which is more abstract than
simple kernel-dives (like, e.g. Mach).

There are many other ways that you can take advantage of this non-polar 
approach to OS design. For example, khttpd is really a fascinating
idea: move only the hardware-sensitive portion of an app into a
module, and then build your application around that core.

Will we see ksmb, kbind, ksmtp and kgnome? I don't know, but this is a 
very interesting time for the OS world, and the future looks to be
even more attractive. Will the release of UnixWare source be useful?
Probably, but I doubt that it will substantally change the decisions
that are being made in light of current OS requirements.

-- 
Aaron Sherman		
ajs@ajs.com		finger ajskey@b5.ajs.com for GPG info. Fingerprint:
www.ajs.com/~ajs	6DC1 F67A B9FB 2FBA D04C  619E FC35 5713 2676 CEAF
	"I want a drug store in The Twilight Zone." -Shriekback



More information about the linux-elitists mailing list