"John S. Gage" wrote:
> Torvald took an existing Unix implementation called Minix and adapted it.
> John Gage
> 
> Joseph Dal Molin wrote:
> > Sorry for what may be a dumb question to many... but how was the kernel for
> > Linux built in the first place.....? Did Mr. T do it by himself.
> >
> > If that was a critical step to moving forward then that's the nut we have to
> > crack.


Gentlemen (aren't there any females BTW?),

please think about loosening up your tight association between
"open source" and "Linux". I have nothing against Linux, but it
opens up the mind if we do not over-constrain our associations.
Linux has helped open source to become popular (= mentioned in the
press), but that doesn't mean Torvald has invented open source.
If anyone has "invented" open source, it is Richard Stallman
with his GNU project and the FSF.  But even Stallman only pointed
something that already existed before him, the spirit of the Unix
community, the sharing of source code, that was both necessary
and possible in the Unix world. 

Sharing of source code was necessary because the different Unices 
weren't binary compatible. Sharing was possible because the APIs of 
these different Unix systems was largely compatible. So, forget about 
the hero-legends (history is not run by singular people), if you want
to understand the open source paradigm, stop staring at Linux, and
understand the origin and spread of Unix at large. 

It starts with Kernighan, the invention of C, the conversion of MULTIX 
to UNIX at Bell Laboratories (today Lucent Technology, Kernighan is 
still there.) Bell Labs (then owned by ATT I think) has issued a source 
code license for the UNIX kernel ... yes, source code license: open 
source!!!

That was back in 1970 or so.  So, it's not wrong to say that Bell
Laboratories "invented" open source!  Yes, the license was expensive,
but was discounted for universities.  UC Berkeley bought in and
they ran operating systems classes and workshops using this code
base. Berkeley added all the rich tools to UNIX, and most importantly
developed the reference IP implementation in 4.2 BSD. Open source
once again made history: most of the worlds IP stacks (commercial
or free) originate in the 4.2 BSD kernel code.  

Now ask yourselves: if Berkeley had not develop IP in BSD and give it 
away as open source, how important would the IP be today?  May be, we 
would still have to live with DEC NET, XEROX XNS, and probably
OSI/ISO would have taken over by now and we were all bogged down by it. 
Would there be e-commerce over the Internet ... who knows, I doubt it. 
But does this make Berkeley the one heroe of open source? No! It's the 
(academic) spirit of giving and taking that open source is all about, not
singular heroes, not ATT, not UCB, not Torvald.

Berkeley UNIX has served as a pooling of all kinds of cool software.
Every Berkeley UNIX had a C compiler, LISP (Franz Lisp) compiler and
interpreter, and of course Fortran.  Lex, yacc, and make were all in 
the box, awk, sed, grep, find, and of course the bourne shell.  Where
would the world be without these little helpers?  You know how it
feels without 'sed' once you had to live with Microsoft DOS and Windows:
poor, powerless. How could you have a software project without make,
SCCI (later RCS used in CVS) and all those things that make up
the UNIX development environment? All this was possible not because
of single heroes or companies or organizations, but only through
the spirit of sharing. So it could all be pooled on the Unix platform.
The BSD platform spanned well-known systems such as the PDP-11 and
the VAX, I don't know why anyone would settle with VMS on the VAX
when BSD was available, but people did.  IS managers did not trust 
the UNIX world without commercial support.  They mostly bet on the
wrong horse, however, as most of the commerial systems they bet on had
to be replaced in the last 10 years, where Unix still lives on.

Anyway, back into the 1980s.  Then came Stallman with Emacs, his 
problems at MIT, his building Emacs again as GNU Emacs, and the GCC 
project.  None of the free Unices nor Linux would exist without GCC.  
Linux is in fact extremely dependent (historically) on the GNU project: 
GCC, GNU make, GNU binutils, bash, GNU shellutils, GNU textutils, 
GNU libc, flex, bison, GNU tar, gzip, etc.  BSD had all this stuff 
already before GNU, but Linux had to take from GNU to keep it's 
stand-alone identity. But all of them, BSD and Linux had to use the 
GCC, since the original Kernighan CC was still ATTs and not shareable.
Stallman is surely a unique character and workhorse, but what would the 
GNU project be without its many supporters, maintaining and writing 
all the code?

In 1989/1990 4.3 BSD was put up on the net, as 4.3 BSD/Net2, open source 
hits again. Then the commercial paradigm hit back: the lawsuit of USL 
(ATT) against Berkeley about copyright infringement with the 4.3 BSD/Net2 
release. Bill and Lynne Jolitz (two more names to be remembered) had already 
delivered the first 4.3 BSD port to the Intel 80386 PC architecture, 
called 386BSD (it's about 1990/91 now.) It was then 1990 that free Unices 
became a real alternative to DOS on the PC. Yes, Tannenbaum's MINIX was 
about first, it costed $100 :-), but it was still somewhat a toy and could 
not accumulate all the existing UNIX tools and applications. 

Linux started out at the same time, and built based on Tannenbaum's work.
If you look at it from the perspective of 10 years ago (1990), where
BSD already was alive and well for over 10 years and 4.3 BSD/Net2 available
for download, Linux was really a superflous one-man-tinker-project. Linux
stayed superflous in its first 5 years, since it had nothing to add that
wasn't already available in the Unix open source world.  What Linux 
magically did, was drawing attention of new users. So, recently you see
crowds of ex-DOSsers and Winsozers who are excited of Linux (but still
a little afraid of Unix.)  So, Linux and 386BSD became quite popular
in about 1993, 1994, but of course far from the mainstream.  The lawsuit 
USL vs. UCB pulled 386BSD back, for fear of the entire BSD being suffocated 
in the lawsuit.  Then Linux' popularity surged, it entered the commercial 
sector, and the rest is a positive feed-back effect that selects a weak 
but distinct enough signal and amplifies it to a big buzz.

Open source is the spirit of sharing, not a designed business paradigm.
There is not one inventor of open source, it is a community phenomenon.  
As always, there are some highlight figures in history, but they live in
a matrix of tradition and fellows.  And there is usually more than one
highlight figure, so in all of history, so in the history of open source.

Why did open source become attractive for the commercial world?  Because
of Linux? Yes and no.  Linux has surged the movement to a level of
attention that was unseen before.  But looking behind the PC magazines,
the open source paradigm was already established in the industry.  Again,
if with anyone specific it started with Richard Stallman and GCC. Stallman's
MANIFESTO said: "free software is about freedom not price" (that's why
"free software" is now called "open source", the shareholders don't like
their company to make just freebies.)  

Apart from consultant firms, who may have operated on the open source 
driven business model for years before, Cygnus was one of the first larger 
companies adopting this model. Cygnus was soon in charge of the GNU C++ 
and then all of the GCC compiler.  Cygnus packed and sold the GNU utilities 
polished nicely and with service. Cygnus finally, finally, made most of 
GNU available for Windows 32 under the name "CygWin". That was the raising 
sun for the poor fellows locked in the DOS/Windows world with its mostly 
proprietory system of vendor and shareware software. Cygnus has been traded
at the stockmarket for quite a while, did fine, was acquired, and continues
to operate, and share.  Yes, RedHat is a hit today with its Linux 
distribution, but its not the first successful open source company.

There is one cornerstone of open source one must not forget: X11.  The X
Window system was maintained in a consortium of all major UNIX vendors.
It was a commercial undertaking, but the source code was free with its
release X11R5.  That allowed people in Germany and Australia (David 
Wexelblat was the guy from Australia) to port X11R5 to i386 Unices. 
Among them was SCO, DELL, and of course 386BSD and Linux.  The Xfree86 
project was born (with nostalgy I can say that I took part in the vote for
the name "Xfree86", althoug my preferred name did not make it :-) That was
back in 1992.  Xfree86 was, since then, maintained by a group in Australia 
(and the world). The control model was, similar as the FreeBSD, a "core 
group" model: a few people, around 10 or so, taking care for the entire 
system, its integrity, and its releases.  Xfree86 was back integrated
into the X11R6 release a few years later.  Without Xfree86, none of the 
free UNIX systems would have created any commercial attention, especially 
not Linux, since Linux was mostly popular in the end-user world.  

One can tell more stories about the X11 project, especially the painful
period under the "Open Software Foundation" (OSF) that tried to do
the impossible: develop open software under a closed source model.
Motif was quite successful and bogged down the open source world, because
back then, most of the more polished GUIs were based on Motif and thus
could not really participate in the open source world. OSF did two
more notable projects: the OSF/1 Unix clone (adopted only by DEC,
now called Tru64 under Compaq) and of course the Distributed Computing
Environment DCE.  DCE was a close forefather of CORBA, one should
know.  But DCE and OSF/1 was not open source, which may be the reason
why DCE really didn't fly all too well. Conversely the DCE competitor,
Sun's ONC RPC with NFS did fly a whole better, because it was open source!

That leads us, finally, to conclude by looking at Sun Microsystems Inc.
Sun is well known in both hardware and operating systems market. Sun
running the Motorola 68000 processor under a 4.2 BSD clone named SunOS. 
But Sun, like every other commercial Unix, developed SunOS under a closed
software model.  That was the reason why upgrading to 4.3 BSD standards
was quite difficult and never really happened.  Finally Sun gave up on
their BSD clone and launched Solaris.  But Sun was not a bad guy in terms
of open source.  They did contribute their RPC protocol along with XDR
and NFS (and NIS, and YP, etc.) to the IETF and gave its source code for 
free.  All RPC and NFS implementations on free Unices (and probably 
everywhere else) are based on the Sun sources.  So, it is only consequential
that something like the Java business model happened.  Java is open
source with some restrictions on what you can do with the sources. Kind
of like the first ATT UNIX source code license, but cheaper. Java and
Sun really have a tradition of good friendship to the open source 
community.

So, this was my essay on a broader picture of what open source is. It 
is the spirit of sharing, the pooling of neat stuff on the Unix environment,
but also the spreading of the idea of sharing accross the boudaries of 
Unix, even including Microsoft Platforms (not because of Microsoft but
because people who are lucky to live outside the MS-only world would share
with those poor fellows who were locked inside.)  Open source is a community 
phenomenon, and its about the fun of writing programs.  So, please don't get 
too focused on Linux, the Linux operational model, Torvald as the hero, the
Copyrignt issue, etc.  Also don't make open source something owned by 
organizations. Open source is a spirit, a spirit that can be organized to 
death, a spirit that can be commercialized to death. Let's just have fun 
writing and sharing programs in source code form.

To finally answer to Joseph:  yes, existing source is the critical step.
A critical mass of software (not necessarily people) is needed. And of course 
users and user demand is also needed (a problem in hospital health care.)  
So, may be, VistA can become such a thing once it gets free of the commercial 
MUMPs.  It seems to me that much of what MUMPS tried to do is all available
on UNIX. The multiprocessing, the files, data bases, etc.  So, a few people's
weightlifting act of dissecting VistA out of the grip of the
MUMPS-and-only-MUMPS
environment could make a big difference. 

regards
-Gunther  

-- 
Gunther_Schadow-------------------------------http://aurora.rg.iupui.edu
Regenstrief Institute for Health Care
1050 Wishard Blvd., Indianapolis IN 46202, Phone: (317) 630 7960
[EMAIL PROTECTED]#include <usual/disclaimer>
begin:vcard 
n:Schadow;Gunther
tel;fax:+1 317 630 6962
tel;home:+1 317 816 0516
tel;work:+1 317 630 7960
x-mozilla-html:FALSE
url:http://aurora.rg.iupui.edu
org:Regenstrief Institute
adr:;;1050 Wishard Blvd;Indianapolis;Indiana;46202;USA
version:2.1
email;internet:[EMAIL PROTECTED]
title:M.D.
fn:Gunther Schadow
end:vcard

Reply via email to