What kinds of things would you want in the GNU OS?

Jim Shankland jas at ernie.Berkeley.EDU
Fri May 26 06:53:53 AEST 1989


In article <10317 at smoke.BRL.MIL> gwyn at brl.arpa (Doug Gwyn) writes:
>In article <106326 at sun.Eng.Sun.COM> bitbug (James Buster) writes:
>>What kinds of things should be in the GNU Kernel?
>
>My opinion is that the GNU kernel should either provide an exact
>duplicate of a standard UNIX system interface, preferably SVR4,
>or it should be a quantum leap forward in OS design....

I vote for the quantum leap forward.  With a little luck and a lot of
skill, arbitrary 4.2, v7, SVR4, etc. UNIX features can be layered on
top as library packages, which makes feeping creaturism less of an evil.

My wish list:  a tiny, but extensible kernel.  A hierarchcal object name
space taking the place of the file system.  Several kinds of objects for
starters:  byte-array like things, including address spaces, disk files,
frame buffers, etc.; data-stream like things, including IPC channels,
terminal devices, etc.; protocol-like things, including both one-input,
one-output things like line disciplines (or anything that's like AT&T
streams), and one-input, many-output (and vice versa) things like TCP,
UDP, etc.; and array multiplexor things like file systems (which take
a large array -- a disk partition -- and split it up into multiple smaller
arrays -- files); and threads.

The code implementing these objects should be freely configurable into the
kernel or into user processes.  Thus, the kernel could end up being tiny,
or gargantuan, depending on how the system is configured.

Much hand waving here, I know.  But I believe there's gold in them thar
hills.  You go find it, though.  Me, I have to pick up my shirts at the
cleaners ....

Jim Shankland
jas at ernie.berkeley.edu

"Blame it on the lies that killed us, blame it on the truth that ran us down"



More information about the Comp.unix.wizards mailing list