ZigZag Ontology - Time to Turn the OS Inside-Out

[Edward Harter <eharter@xxxxxxxxx>]

As the Apple Web page about MkLinux says,

"Operating system elitists (read, some kernel hackers) may disagree, but
the rest of us know a kernel, however wonderful, isn't enough. We need
more: shells, utilities, a window system, and all those other little toys
we've grown to love. Don't worry; MkLinux has everything you've come to
love in Linux. "

For some 5 years a debate raged - fruitlessly - as to whether Windoze in
its 3.x incarnation was an OS: of course, for the hardware-driver fixated
types, Wxxdoze was anything _but_ an OS, and on a purely technical basis
they were right. But somehow, on the path to OS purity and perfection,
Micro$oft managed to turn it's GUI/windowing shell (feel free to substitute
any combination noun-verb hyphenated pair here that will make you feel
good;-) into an OS - at least I haven't heard anyone complain, criticize,
or even bring up the topic since Win95 rolled over the world, with only
vestigial elements from its crude, DOS-bound roots discernible in its
deeper layers.

Ah, you may well ask, how is this relevant? Well, back in the days of the
Dinosaurs (pre-PDP really, but in practical terms until the Apple II,
Commodore 64, and IBM PC) what _mattered_ about an OS was how efficiently
the T. Rex (IBM 360) and Brontosaurs (370 and the like) were made use of -
humans were there simply to shovel in stacks of cards, and deal with the
mountainous paper excreta of these behemoths. Jumping forward to the
mid-80's, when Bill Gates stole an idea that had already been stolen, and
incarnated the truly absurd and unusable Windows 1.0, you can be sure he
was planning - in his opportunistic way - not only to hold off Apple's GUI,
the Mac, but eventually to turn it into an OS when the time was right. And
in the space of a decade or less, he managed to do just that.

The ultimate point here revolves not around whether a particular piece of
software performs one or more of the functions which are considered to
define an OS, but rather around the fact that the economics and sociology
of computing have increasingly, year by year, moved to the human side and
away from pure hardware considerations. In fact, if you look at certain
developments in microkernels, whether in Mach or even Windows NT, or the
so-called Rhapsody design concept, you will see that the software layer
most closely bound to hardware appears more and more to be handled by the
microkernel, upon which any number of OS's could happily sit. And this is
very clearly and explicitly by _design choice_ of the University
researchers who did the original microkernel work. Eg, Mach.

Thus, hypothetically, if Ted were to ask that ZZ incorporate a few modules
to let it speak to the microkernel it rides upon, potentially
1. it would be an OS, should it have that (large) goal in mind
2. it could do away with _everything else_ in that given system except
itself, its data, and the minimal interface elements it would need to
function, along with the above-mentioned system calls.

Again, remember how minimalist Unix was in the days of Dennis Ritchie and
Ken Thompson: some PDP assembler routines; bootstrap C; and the OS 'stuff'
riding on top of that, coded in a mixture of assembler and C. Plus a few
apps to play with, but not many to begin with.

Another way of stating it: if you could port a GNU C compiler _into_
Zigzag, conceptually speaking, the job would be done. Practically speaking,
you'd still have a small mountain of man-years to make all the tools and
toys people would want - but if you made the compile process as easy to use
as the Mac interface is to open an app and print a page, _you wouldn't need
to compile anything_.. Let the user pull on the joint, take a deep breath,
and by the time he breathes out, Hey Presto! magic instant apps. (Ok, for
you guys who don't get it, this is how easy and simple the app _creation_
process is supposed to be, via ZigZag/joint compile).

Happy Hunting,

ed harter (aka Captain Midnight)