17 Jan
2020
17 Jan
'20
4:01 p.m.
[I originally asked the following on Twitter which was probably not the smartest idea]
I was recently wondering about the origins of Linux, i.e. Linux Torvalds doing his MSc and
deciding to write Linux (the kernel) for the i386 because Minix did not support the i386
properly. While this is perfectly understandable I was trying to understand why, as he was
in academia, he did not decide to write a “free X” for a different X. The example I picked
was Plan 9, simply because I always liked it but X could be any number of other operating
systems which he would have been exposed to in academia. This all started in my mind
because I was thinking about my friends who were CompSci university students with me at
the time and they were into all sorts of esoteric stuff like Miranda-based operating
systems, building a complete interface builder for X11 on SunOS including sparkly mouse
pointers, etc. (I guess you could define it as “the usual frivolous MSc projects”) and
comparing their choices with Linus’.
The answers I got varied from “the world needed a free Unix and BSD was embroiled in the
AT&T lawsuit at the time” to “Plan 9 also had a restrictive license” (to the latter my
response was that “so did Unix and that’s why Linus built Linux!”) but I don’t feel any of
the answers addressed my underlying question as to what was wrong in the exposure to other
operating systems which made Unix the choice?
Personally I feel that if we had a distributed OS now instead of Linux we’d be better off
with the current architecture of the world so I am sad that "Linux is not Plan
9" which is what prompted the question.
Obviously I am most grateful for being able to boot the Mathematics department’s MS-DOS
i486 machines with Linux 0.12 floppy disks and not having to code Fortran 77 in Notepad
followed by eventually taking over the department with X-Terminals based on Linux
connected to the departmental servers (Sun, DEC Alpha, IBM RS/6000s). Without Linux they
had been running eXeed (sp?) on Windows 3.11! In this respect Linux definitely filled in a
huge gap.
Arrigo
17 Jan
17 Jan
4:53 p.m.
On Fri, Jan 17, 2020 at 9:42 AM Arrigo Triulzi <arrigo(a)alchemistowl.org>
wrote:
[I originally asked the following on Twitter which was
probably not the
smartest idea]
I was recently wondering about the origins of Linux, i.e. Linux Torvalds
doing his MSc and deciding to write Linux (the kernel) for the i386 because
Minix did not support the i386 properly. While this is perfectly
understandable I was trying to understand why, as he was in academia, he
did not decide to write a “free X” for a different X. The example I picked
was Plan 9, simply because I always liked it but X could be any number of
other operating systems which he would have been exposed to in academia.
This all started in my mind because I was thinking about my friends who
were CompSci university students with me at the time and they were into all
sorts of esoteric stuff like Miranda-based operating systems, building a
complete interface builder for X11 on SunOS including sparkly mouse
pointers, etc. (I guess you could define it as “the usual frivolous MSc
projects”) and comparing their choices with Linus’.
The answers I got varied from “the world needed a free Unix and BSD was
embroiled in the AT&T lawsuit at the time” to “Plan 9 also had a
restrictive license” (to the latter my response was that “so did Unix and
that’s why Linus built Linux!”) but I don’t feel any of the answers
addressed my underlying question as to what was wrong in the exposure to
other operating systems which made Unix the choice?
The AT&T lawsuit (April 1992) post-dated Linus starting on his work (eg
0.12 released January 1992). He said in an interview once he was unaware
that net/2 was out and could be leveraged to get a working system when he
started. It did give a big boost to Linux at a critical time due to the
huge amount of FUD that it created over BSD's future.
Warner
5:08 p.m.
On 17 Jan 2020, at 17:53, Warner Losh <imp(a)bsdimp.com> wrote:
The AT&T lawsuit (April 1992) post-dated Linus
starting on his work (eg 0.12 released January 1992). He said in an interview once he was
unaware that net/2 was out and could be leveraged to get a working system when he started.
It did give a big boost to Linux at a critical time due to the huge amount of FUD that it
created over BSD's future.
Well, that plainly removes the AT&T lawsuit argument from the answers list so the
remaining question I had about its origins is clearer.
5:25 p.m.
Since I was a close observer and an early adopter of Plan 9, which is still my main
development platform, I'll contribute my thoughts on the question.
Plan 9 was still being developed in the time frame that Torvalds was working out Linux.
Everyone working in the Unix field wanted a version that was not under AT&T's
expensive license. The last one I bought was north of $64K.
The biggest reason that Plan 9 isn't the current Linux is that the world didn't
need a cloud operating system in the early 1990's. They just needed a Unix for which
they could get the source.
Plan 9 solves the problem of "How do I make a bunch of machines look like a single
system?" If you wanted to mess around with a system in the early 1990's you
didn't have a bunch of people and a bunch of systems you needed to make appear as
one. You just had a single box.
So, my Plan 9 remains small. In fact, I've been removing things from it, like local
disks, that is contrary to the original vision. (Or set of visions. I remember getting a
lot of different answers form everyone involved in 1127 about what it was that they were
doing.)
The general question of why Linux emerged and not others didn't is a very hard
question that computer historians will be researching for a lot time. It's complex,
like all economics.
Brantley
On Jan 17, 2020, at 11:01 AM, Arrigo Triulzi
<arrigo(a)alchemistowl.org> wrote:
[I originally asked the following on Twitter which was probably not the smartest idea]
I was recently wondering about the origins of Linux, i.e. Linux Torvalds doing his MSc
and deciding to write Linux (the kernel) for the i386 because Minix did not support the
i386 properly. While this is perfectly understandable I was trying to understand why, as
he was in academia, he did not decide to write a “free X” for a different X. The example I
picked was Plan 9, simply because I always liked it but X could be any number of other
operating systems which he would have been exposed to in academia. This all started in my
mind because I was thinking about my friends who were CompSci university students with me
at the time and they were into all sorts of esoteric stuff like Miranda-based operating
systems, building a complete interface builder for X11 on SunOS including sparkly mouse
pointers, etc. (I guess you could define it as “the usual frivolous MSc projects”) and
comparing their choices with Linus’.
The answers I got varied from “the world needed a free Unix and BSD was embroiled in the
AT&T lawsuit at the time” to “Plan 9 also had a restrictive license” (to the latter my
response was that “so did Unix and that’s why Linus built Linux!”) but I don’t feel any of
the answers addressed my underlying question as to what was wrong in the exposure to other
operating systems which made Unix the choice?
Personally I feel that if we had a distributed OS now instead of Linux we’d be better off
with the current architecture of the world so I am sad that "Linux is not Plan
9" which is what prompted the question.
Obviously I am most grateful for being able to boot the Mathematics department’s MS-DOS
i486 machines with Linux 0.12 floppy disks and not having to code Fortran 77 in Notepad
followed by eventually taking over the department with X-Terminals based on Linux
connected to the departmental servers (Sun, DEC Alpha, IBM RS/6000s). Without Linux they
had been running eXeed (sp?) on Windows 3.11! In this respect Linux definitely filled in a
huge gap.
Arrigo
7:59 p.m.
On Fri, Jan 17, 2020 at 05:01:47PM +0100, Arrigo Triulzi wrote:
I was recently wondering about the origins of Linux,
i.e. Linux Torvalds
doing his MSc and deciding to write Linux (the kernel) for the i386
because Minix did not support the i386 properly.
If I remember correctly from those days as a student, that was not the
starting point for Linux.
He did not 'decide' to write Linux at the start.. He initially didn't even
decide to write an OS at all..
As I recall he actually got an i386 based machine and wanted to explore
the features of that CPU compared to the 286 and the like. To do that he
decided to basically write (if I remember right..) an editor that ran
directly on the hardware and made use of the new i386 features.
But of course that meant he had to do his own filesystem code to read/write
stuff, do some sort of memory management, have some sort of internal
'task' scheduler, etc.
At some point, in hindsight probably crucial, he came to the conclusion that
it was starting to look more like an OS kernel and looked for something
that he could fairly easily run the userland binaries from.
MINIX was the obvious one as a userland 'donor' here. Code AND installation
media were easily available on discs to us in europe at the time, unlike
many of the others out there.
At the time Linus was in Finland and most European universities and colleges
in those days (late 80's start 90's) were very, very wary of any legal
implications even before all the lawsuits. So as a result anything with a
big license text on it (even 'kinda free' ones like BSD) were classed as
'risky' and access kept to a bare minimum.
Add fledgeling internet access that was also highly restricted because the
telco costs in europe were usually massively higher than in the US, so
learning about new stuff was harder and then getting it from 'somewhere'
was often a painful process.
At the time I suspect he just wanted to 'finish' the little i386
learing project with a 'look what I made!' and move on once he did that.
So chasing after some other, more esoteric OS, just wasn't worth the effort
at the time.
However, once he did publish it the MINIX userland basically became the
'incubator' for this new little (monolithic! :P ) kernel broke out when
GCC and slowly the other GNU tools became available so it could self-host.
Somehow Linus at that time found so much fun from getting all these
patches and code to stick into his funny little kernel and watching it
grow and evolve that he stuck with it and didn't move to other projects.
So all in all.. As I remember it, there was never really a decision to 'make
this great new OS!'.. It kinda happened with right place, right time, right
people, etc.
I vaguely remeber that Linus did give such a timeline in an interview once..
Bye, Arno.
18 Jan
18 Jan
3:50 a.m.
On Fri, Jan 17, 2020 at 08:59:08PM +0100, Arno Griffioen wrote:
If I remember correctly from those days as a student,
that was not the
starting point for Linux.
He did not 'decide' to write Linux at the start.. He initially didn't
even
decide to write an OS at all..
As I recall he actually got an i386 based machine and wanted to explore
the features of that CPU compared to the 286 and the like. To do that he
decided to basically write (if I remember right..) an editor that ran
directly on the hardware and made use of the new i386 features.
Not an editor, but rather a terminal emulator, to talk to a modem so
he could connect to the university's computers.
He first started by having two processes, one which would print just
'A' characters, and another that would print 'B', and had a simple
scheduler so they would alternate printing to the screen: "AAAABBBB",
etc., then "ABABABAB", etc. And from there he wrote the terminal emulator.
What he did after that isn't well recorded; it could have been an
editor, but for a long time he using Minix for editing and compiling
the proto-kernel that would later become Linux.
However, once he did publish it the MINIX userland
basically became the
'incubator' for this new little (monolithic! :P ) kernel broke out when
GCC and slowly the other GNU tools became available so it could self-host.
Somehow Linus at that time found so much fun from getting all these
patches and code to stick into his funny little kernel and watching it
grow and evolve that he stuck with it and didn't move to other projects.
At the time when Linus announced his creation (not yet named) on
comp.os.minix in August 1991, it was already self-hosting. And that
happened pretty quickly; he first started working on the project in
June or July.
Around the end of 1991, I had added Job Control (implemented from
POSIX.1 as a the specification), so we could put jobs in the
background. In 1992 X Windows was ported to Linux. Networking
support followed shortly thereafter.
So all in all.. As I remember it, there was never
really a decision to 'make
this great new OS!'.. It kinda happened with right place, right time, right
people, etc.
In the super-early days (late 1991, early 1992), those of us who
worked on it just wanted a "something Unix-like" that we could run at
home (my first computer was a 40 MHz 386 with 16 MB of memory). This
was before the AT&T/BSD Lawsuit (which was in 1992) and while Jolitz
may have been demonstrating 386BSD in private, I was certainly never
aware of it --- and Linus was publishing new versions every few days
on an ftp site. We'd send patches, and in less than a week, there'd
be a new release dropped that we could download.
So the argument, "Linus would have never started on Linux if itT
weren't for the AT&T Lawsuit" I don't think fits with the timeline.
Development was very fast paced, and so it was *fun*. And at least
for me, the lacking of networking during the early days didn't bother
me much, since I didn't have networking at home. (I didn't have
grounded outlets, either, in my 3 people for $1050/month apartment.
Each leg was 50-60V to ground, and the wiring was cloth wrapped, and
was either steel or aluminum; I never did figured out which....)
Using zmodem over a 2400 bps modem was way more efficient than PPP, so
even once we had networking, I didn't always bring up pppd. And the
most common way I would download source was using set of 1.44 MB
floppies and a station wagon (literally; I was driving a Corolla wagon).
During those early days, the fact that Linux was more "primitive" than
BSD may have been an advantage, since it sources was small, and
release engineering is simple when you only support one architecture.
The other things I noticed was that because we didn't have the weight
of the Unix/BSD legacy, we were more free to experiment. Bruce Evans
was working on the serial driver for FreeBSD, and I was working on the
serial driver for Linux, and we had a friendly competition to see who
could get better throughput using the very primitive 8250 and later
16550 UART. The figure of merit we were using was the CPU overhead of
a C-Kermit file transfer over two RS-232 ports connected via a
loopback cable. We'd compare notes to see how we could make things
better, me for Linux, and Bruce for FreeBSD, and it was *fun*.
Eventually, it got to the point where I was making changes to the tty
layer to further optimize things, and at that point Bruce reported
that he couldn't do some of the optimizations, since it would have
required changing the TTY layer that had been handed down from the
Gods of Olympus^H^H^H^H^H^H^H^H BSD and so it was nixed by his
colleagues in FreeBSD land.
In contrast, in Linux, people felt free to rip out and replace code if
it would make things better. Depending on how you count things, the
networking layer in Linux was ripped out and replaced three or four
times in the space of as many years. Sure, the first version was
pretty crappy, and was barely good enough for simple telnet
connections. But things got better fast, because people were felt
free to experiment.
My personal belief is that it was this development velocity and
freedom to experiment starting with a super simple base is what caused
Linux to become very popular amongst the those who just wanted to play
with kernel development. Compare and contrast Linus's willingness to
accept patches from others and his turnaround time to get those
patches into new releases with Bill Jolitz's 386BSD effort --- and I
don't think you need the AT&T lawsuit to explain why Linux took off in
1991-1992. FreeBSD and NetBSD was started in 1993 because of the
failure of Jolitz to accept patches in a timely fashion.
- Ted
4:19 a.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
On Friday, 17 January 2020 at 22:50:51 -0500, Theodore Y. Ts'o wrote:
In the super-early days (late 1991, early 1992), those of us who
worked on it just wanted a "something Unix-like" that we could run at
home (my first computer was a 40 MHz 386 with 16 MB of memory). This
was before the AT&T/BSD Lawsuit (which was in 1992) and while Jolitz
may have been demonstrating 386BSD in private, I was certainly never
aware of it
At the start of this time, Bill was working for BSDI, who were
preparing a commercial product that (in March 1992) became BSD/386.
As Rob Kolstad told me in mid-1992, he had a (to put it mildly)
difference of opinion with Rob, and probably others, about the
commercial nature (at the time BSD/386 cost $1000), and he resigned at
the end of 1991. So 386BSD presumably didn't exist before that.
On the other hand, Bill did write the articles in Dr. Dobbs Journal,
which started in January 1991, so my best guess is that Linus just
wasn't informed about the developments. It's easy to forget how
difficult it was to get this kind of information in those days. I was
informed about the articles, but more by chance than anything else. I
didn't find out about BSD/386 until March 1992, when the first betas
were released.
Greg
--
Sent from my desktop computer.
Finger grog(a)lemis.com for PGP public key.
See complete headers for address and phone numbers.
This message is digitally signed. If your Microsoft mail program
reports problems, please read http://lemis.com/broken-MUA
3:25 p.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
On Sat, Jan 18, 2020 at 03:19:13PM +1100, Greg 'groggy' Lehey wrote:
On Friday, 17 January 2020 at 22:50:51 -0500, Theodore
Y. Ts'o wrote:
Wikipedia says he was working on 386BSD as early has 1989 and that
clicks with me (Jolitz worked for me around 1992 or 3). I don't
remember him mentioning working at BSDI, are you sure about that
part? Those guys did not like each other at all.
In the super-early days (late 1991, early 1992), those of us who
worked on it just wanted a "something Unix-like" that we could run at
home (my first computer was a 40 MHz 386 with 16 MB of memory). This
was before the AT&T/BSD Lawsuit (which was in 1992) and while Jolitz
may have been demonstrating 386BSD in private, I was certainly never
aware of it
At the start of this time, Bill was working for BSDI, who were
preparing a commercial product that (in March 1992) became BSD/386. 
4:19 p.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
I don't remember him mentioning working at BSDI,
are you sure about
that part?
See https://tech-insider.org/unix/research/1992/0319.html
coworkers said:
"Jolitz was one of the founders of Berkeley Software Design, Inc. ...
was a full time employee of BSDI for 11 months of 1991 -- from
January 1, 1991 through November 30, 1991
... a founder of BSDI as well as its first full-time paid employee"
And his wife wrote "Bill did work for UUNET from January to
June of 1991."
Jeremy C. Reed
echo Ohl zl obbx uggc://errqzrqvn.arg/obbxf/csfrafr/ | \
tr "Onoqrsuvxzabcefghl" "Babdefhikmnoprstuy"
19 Jan
19 Jan
2:49 a.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
On Sat, Jan 18, 2020 at 03:19:13PM +1100, Greg 'groggy' Lehey wrote:
On the other hand, Bill did write the articles in Dr.
Dobbs Journal,
which started in January 1991, so my best guess is that Linus just
wasn't informed about the developments.
I don't believe that to be correct. "Porting Unix to the 386: Missing
Pieces, Part 1", by William Frederick Jolitz and Lynne Greer Jolitz,
was published in the May 1992 issue of Dr. Dobbs Journal.
https://www.drdobbs.com/architecture-and-design/porting-unix-to-the-386-mis…
- Ted
3:12 a.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
On Saturday, 18 January 2020 at 21:49:00 -0500, Theodore Y. Ts'o wrote:
On Sat, Jan 18, 2020 at 03:19:13PM +1100, Greg
'groggy' Lehey wrote:
That appears to be a different pair of articles. My quotation was
from Wikipedia, which I believe to be correct I was working in the USA
in those days, and I saw the magazine in a bookshop in Austin TX. I
stopped working there in December 1991. Unfortunately I wasn't very
interested at the time ("who cares about BSD?"). Check the references
in the Wikipedia article. Unfortunately the Dr. Dobbs site can't find
them.
May 1992 was significantly after the first release of 386BSD in March
1992.
Greg
--
Sent from my desktop computer.
Finger grog(a)lemis.com for PGP public key.
See complete headers for address and phone numbers.
This message is digitally signed. If your Microsoft mail program
reports problems, please read http://lemis.com/broken-MUA
On the other hand, Bill did write the articles in
Dr. Dobbs Journal,
which started in January 1991, so my best guess is that Linus just
wasn't informed about the developments.
I don't believe that to be correct. "Porting Unix to the 386: Missing
Pieces, Part 1", by William Frederick Jolitz and Lynne Greer Jolitz,
was published in the May 1992 issue of Dr. Dobbs Journal.
https://www.drdobbs.com/architecture-and-design/porting-unix-to-the-386-mis…
3:47 a.m.
New subject: Early Linux and BSD
On Sun, Jan 19, 2020 at 02:12:25PM +1100, Greg 'groggy' Lehey wrote:
May 1992 was significantly after the first release of
386BSD in March
1992.
https://dl.acm.org/doi/abs/10.5555/119491.121544 shows:
Dr. Dobb's Journal Vol. 16, No. 8
"The basic kernel: overview and initialization"
Authors: William Frederick Jolitz, Lynne Greer Jolitz
Publication: Dr. Dobb's Journal June 1991
Cheers, Warren
3:51 a.m.
New subject: Early Linux and BSD
On Sunday, 19 January 2020 at 13:47:21 +1000, Warren Toomey wrote:
On Sun, Jan 19, 2020 at 02:12:25PM +1100, Greg
'groggy' Lehey wrote:
Yes, that's one of the articles referenced in the references on the WP
page, specifically
https://en.wikipedia.org/wiki/386BSD#Further_reading. This one was
the 9th of 18 articles, which ran from January 1991 to July 1992.
Greg
--
Sent from my desktop computer.
Finger grog(a)lemis.com for PGP public key.
See complete headers for address and phone numbers.
This message is digitally signed. If your Microsoft mail program
reports problems, please read http://lemis.com/broken-MUA
May 1992 was significantly after the first
release of 386BSD in March
1992.
https://dl.acm.org/doi/abs/10.5555/119491.121544 shows:
Dr. Dobb's Journal Vol. 16, No. 8
"The basic kernel: overview and initialization"
Authors: William Frederick Jolitz, Lynne Greer Jolitz
Publication: Dr. Dobb's Journal June 1991
3:58 a.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
On Sunday, 19 January 2020 at 14:12:25 +1100, Greg 'groggy' Lehey wrote:
On Saturday, 18 January 2020 at 21:49:00 -0500,
Theodore Y. Ts'o wrote:
Sorry, I was wrong. That particular article was number 16 of the 18
part series, as shown at
https://en.wikipedia.org/wiki/386BSD#Further_reading. It does raise
the question why the Dr Dobb's search engine didn't find any of them.
Greg
--
Sent from my desktop computer.
Finger grog(a)lemis.com for PGP public key.
See complete headers for address and phone numbers.
This message is digitally signed. If your Microsoft mail program
reports problems, please read http://lemis.com/broken-MUA
I don't believe that to be correct. "Porting Unix to the 386: Missing
Pieces, Part 1", by William Frederick Jolitz and Lynne Greer Jolitz,
was published in the May 1992 issue of Dr. Dobbs Journal.
https://www.drdobbs.com/architecture-and-design/porting-unix-to-the-386-mis…
That appears to be a different pair of articles.
1:25 p.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
On Sun, Jan 19, 2020 at 02:58:08PM +1100, Greg 'groggy' Lehey wrote:
Sorry, I was wrong. That particular article was number 16 of the 18
part series, as shown at
https://en.wikipedia.org/wiki/386BSD#Further_reading. It does raise
the question why the Dr Dobb's search engine didn't find any of them.
Out of curiosity, did the articles contain download information for a
bootable copy of 386BSD?
- Ted
1:48 p.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
I don't remember, to be honest, I don't think so. When it first was
released which was before Linus's famous announcement, the ftp site was
from ucbvax and any UCB licensee that asked for got sent a copy of the URL
but you had to ask for it.
It was not a well kept secret. When the BSDi / UCB vs ATT came later, as
Larry said a lot of us who were primarily driven by wanting a 'real unix'
for the 386 thought it was about copyright and that's when we looked at
Linux. Then it came out that was a TS suit and many of us switched back
because Linux lacked networking and X11
On Sun, Jan 19, 2020 at 8:26 AM Theodore Y. Ts'o <tytso(a)mit.edu> wrote:
On Sun, Jan 19, 2020 at 02:58:08PM +1100, Greg
'groggy' Lehey wrote:
--
Sent from a handheld expect more typos than usual
Sorry, I was wrong. That particular article was number 16 of the 18
part series, as shown at
https://en.wikipedia.org/wiki/386BSD#Further_reading. It does raise
the question why the Dr Dobb's search engine didn't find any of them.
Out of curiosity, did the articles contain download information for a
bootable copy of 386BSD?
- Ted
20 Jan
20 Jan
3:32 a.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
At Sun, 19 Jan 2020 08:25:51 -0500, "Theodore Y. Ts'o"
<tytso(a)mit.edu> wrote:
Subject: Re: [TUHS] Early Linux and BSD (was: On the origins of Linux - "an academic
question")
On Sun, Jan 19, 2020 at 02:58:08PM +1100, Greg 'groggy' Lehey wrote:
Yes, they did:
https://www.drdobbs.com/porting-unix-to-the-386-the-final-step/184408800
See also:
http://gunkies.org/wiki/386BSD
Also keep in mind that NetBSD started as a set of "net" (as in usenet)
patch kits for 386BSD.
--
Greg A. Woods <gwoods(a)acm.org>
Kelowna, BC +1 250 762-7675 RoboHack <woods(a)robohack.ca>
Planix, Inc. <woods(a)planix.com> Avoncote Farms <woods(a)avoncote.ca>
Sorry, I was wrong. That particular article was number 16 of the 18
part series, as shown at
https://en.wikipedia.org/wiki/386BSD#Further_reading. It does raise
the question why the Dr Dobb's search engine didn't find any of them.
Out of curiosity, did the articles contain download information for a
bootable copy of 386BSD? 
3:51 a.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
It does me no credit, that I initially reacted very badly to 386BSD,
and the initial {Net,Free,Open} situation.
I found all this "fragmentation" pretty hard to understand. -BSDI felt
like it had occupied the space, and I couldn't entirely understand
what was going on, or why any of it mattered.
I also reacted very badly to the public dirty linen spats around the
Jolitz's IPR, and who-did-what to-whom in woodshed, with colonel
mustard and the lead pipe.
What worried me was the loss of intellectual capital. People I liked
online, and respected, seemed to be running to the four winds. Why was
something I worked with turning to suck?
What I think I missed (didn't understand) was how draining support was
for Berkeley, and in the absence of a sugar daddy (Earth Sciences?
DOE? IBM?) and loss of contracts like the NSFNet support for BSD-rt
(they stopped using PC-Rt as a platform for routing) It was
increasingly hard for a teaching and research institution to justify
what was going on. Sun spun out of Stanford. MIT was doing the Gui
work. Compilers had gone really funny with Gnu and an income stream
had evaporated, and a lawsuit was in the offing, and people who hadn't
done full on DOTCOM boom vesting suddenly found growing old and not
having a 401k at the scale they needed to maintain a wine cellar ...
What I also missed is that it stopped innovating. People were
innovating in other places, doing things I didn't understand. I
totally did NOT get what 8th ed. and Plan9 was on. Drugs I couldn't
fathom.
Nothing lasts forever. Had you said to me "ha.. VMS is going to die,
and Dec is going to die, and Sun is going to fold into Larry Ellisons
personal empire" I'd have taken every one of those bets on what turned
out to be the loosing side.
3:59 a.m.
New subject: Early Linux and BSD
On 1/19/2020 7:51 PM, George Michaelson wrote:
What I think I missed (didn't understand) was how
draining support was
for Berkeley
Two of the founders of NetBSD were undergrads who were also members of
the Postgres research group that I was a part of. They did
NetBSD without any formal support from UC Berkeley. Whether or not
this was a good thing is debatable.
Jon
5:19 p.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
On Sun, Jan 19, 2020 at 10:52 PM George Michaelson <ggm(a)algebras.org> wrote:
It does me no credit, that I initially reacted very
badly to 386BSD,
and the initial {Net,Free,Open} situation.
First, be careful. What we sometimes call 386BSD as a 'release'
started
just as a port of NET2 to the 386 based 'commodity' hardware platform. The
history is that in the late 1970s/early 80s Bill Jolitz was working for Nat
Semi and ported BSD 4.1, to a multibus based NS16032 board that NS had
built, which was similar to the Stanford University Network (SUN) terminal
what had a 68000. He eventually built a 'luggable' using that and updated
to the port to 4.2++. He (and Lynn I believe) started a company to sell
that hardware/software solution and for whatever reason, it did not really
take off.
At some point, he got his hands on a 386based PC (Compaq I think) and
started to move his port over to that system. A number of people helped
him (for instance I did a bunch of the AT/disk controller work as I had
access to the WD design documents for another consulting gig I had at the
time - Bill mentioned this in the articles BTW).
Bill and Lynn's NS16032 and 386 code went back to the CSRG 'masters' -
although how and that happened was never completely clear to me. The SCCS
deltas tell at least part of the story. Bill managed to make a bootable
image that mostly installed on a PC/386 as the minicomputer versions did
from the formal release. The ftp area of ucbvax had all of these bootable
images available for download such as one for an HP 68K system and I think
the DEC VAX and PMAX, the CCG system and a few others IIRC. As I have said
in other messages if you were a UCB licensee you had the passwords to
look/download from that area. Bill placed that version in the same ftp
area. The 386 based port went viral at least with the UCB licensees. (In
fact, if Linus had known about it, theoretically he could have used it
also. His university was licensee, but as Larry McVoy likes to point, not
all the schools were as free with the IP, so I will not go down that
rathole).
The bottom line is that many people (like me on a Wyse386) started with
Bill's original port; including the BSDi founders.
When Jolitz and BSDi went separate ways, Jolitiz continued to update the
CSRG 386 based tarball (to an extent). One of the issues was there
originally was attempt to keep the different architectural versions of BSD
in sync ( to a point and NetBSD does yet exist). A number of people were
unhappy and the speed, depth *etc*. of the 386 version, most notably Jordan
Hubbard and FreeBSD was born. The two biggest issues Jordan wanted to fix,
was easier install and a bit wider support for more hardware (again I sent
Jordan the changes to FreeBSN 1x for the Wyse and a couple of NCR boxes).
The NetBSD project would birth from the original ideals of CSRG and trying
to keep everything the same but that's still in the future.
I found all this "fragmentation" pretty hard to understand. -BSDI felt
like it had occupied the space, and I couldn't entirely understand
what was going on, or why any of it mattered.
See below....
What I think I missed (didn't understand) was how
draining support was
for Berkeley, and in the absence of a sugar daddy
Herein is the issue that many people on the sidelines missed.
CRSG was a large project and funded a lot of work at UCB in EECS. It never
funded me (I was funded by Tektronix, HP, DEC *et al*), but that project
did a fund a number of students. However, at some point CSRG stopped being
a research project and started being a support project for DARPA. There
was also a good deal of resentment by some groups in EECS that were not
getting DARPA funding.
I'll not say if that was good or bad but I will say that it did cause great
deal consternation at UCB within the department and many people doing more
formal research were not happy. In the end, the EECS Department mothers
and fathers along with the Dean *et al*, decided to stop/end the CSRG
project. Many people who were directly or indirectly working on BSD, like
Mary Ann and myself, had graduated and had since left. Bob Kridle had
formed Mt. Xinu, Asa Romberger has formed Unisoft, Joy had left/was leaving
for Sun, *etc*. So the question remained what to do with CSRG. As to
what everyone would do, became every person for her/himself and as we know
some of the folks, along with a few folks from the USENIX community formed
BSDi.
As was noted elsewhere, NetBSD would eventually be formed by volunteers to
keep the different ports alive (in fact much of the efforts was from folks
not at UCB), but that was still in the offing. Remember, while CSRG
itself was not a research project, a lot of people around the world were
using the BSD code base for their own research. The whole idea of NetBSD
was to create a uniform platform that people could compare things. So, the
question of how that was to come about or do any work on BSD if DARPA was
not paying the bills, was still an open one. But, the idea that would
eventually create FreeBSD, was supporting a pure commodity *solution for
day-to-day use, not as a research platform*. [I'll leave off the later
OpenBSD/NetBSD fork by Theo here as it has little to do with the question].
BSDi had a similar/same goal of producing something like SunOS/VMS *etc* but
supported on commodity hardware. That solution was to sell it and using
the revenues from the support contract, be able to pay people to do that
work. As I said and in some other messages, it is noted that Bill Jolitz
wanted something more FOSS. Truth is BSDi code was 'open source' but it
took a $1K license to *get the source from them*.
In the end, the real problem was not the infighting between the different
BSD camps, but AT&T, who wanted the entire pie. Clearly, their executives
saw anything other than their complete control of the UNIX IP as a threat.
Hence the court case, the eventually AT&T/Sun relationship *etc*...
Your lack of 'sugar daddy,' really comes back to that. There were few
people at the time that could pay the bills. Until then DARPA had been
it. I do not know if DARPA wanted out or if another group could have been
formed that could take over CSRG. I did have discussions with Rob over a
beer that at least the thought had crossed the BSDi folks mind, that once
started; they would apply for a DARPA contract.
At the time had blow up, I was a consultant and I personally was
considering what I was going to do next and if they had had a real future,
the talks with Rob might have gotten more serious. My wife wanted me to
stop being independent if we were to start a family (I would join Locus
instead).
BTW: I was in an interesting position as I was friends with all of the
different sides in the war/original fight. Like Jolitz, I wanted to see
what we now call a 'FOSS' release of BSD. But like Rob, I knew it was
going to take some revenue stream to make it happen/continue the support.
In the end, the AT&T legal mess blew it all up. BSDi ended up failing
and Jordan's work stayed around.
BTW: what pays for Linux development these days by number of 'committers
salary' is Intel (#1), IBM (#2), then a load of other firms including the
different distros. But for *any* platform to be successful and actually
continue to be used in the market, someone has to pay the salaries of some
set of professional programmers to do the work.
That said when AT&T injoined BSDi and UCB a lot of people (myself included)
started to hack on Linux. But just think if AT&T had actually won the case
and courts decided UNIX was allowed to be a trade secret, then Linux and
all of the UNIX 'clones' would have been in violation.
No matter what flavor of UNIX you like, we are all in debt to UCB and BSDi
for settling the IP argument. The court was clear, the >>ideas<< behind
UNIX (*a.k.a.* the intellectual property) came from Ken, Dennis and friends
at AT&T and *they did own it.* But because of the 1956 consent decree
that published the ideas and the moment the ideas were published, we all
can now >>use<< them. The provenance of the source code does not relate to
the provenance of the idea, so* the source code itself does not define what
UNIX is or is not. *
I bring this all up in hopes to try to close this rat hole of Linux, *vs*.
*BSD. Like editors, we all have our own favorites. That's cool, we don't
want one thing to be forced down our throat. Having a choice is what is
good. And what I value, Larry or Jon may not necessarily like. Most of
us if not all on this list probably want something that approximates Ken
and Dennis's original ideas not what IBM, DEC, CDC were trying to make us
use in the old days or what Microsoft calls a system today.
The discussion of how we got there and what people valued at the time is
useful so we can try to remember the history and learn from it; but getting
into right/wrong, good/bad, or you could have had this is a tad tiresome;
IMO.
5:49 p.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
Thanks Clem.
One minor clarification. Jordan and the patchkit work did predate NetBSD.
However, the NetBSD project formed a little before the FreeBSD project that
grew out of the patchkit days. Jordan didn't get that moving until NetBSD
made rumblings... it was still a time that you heard a lot of what was
going on by word of mouth, not so much by postings and email...
The OpenBSD split was years later... and a complicated mix of personality
conflicts and technical differences. But in many ways it was a smaller
split since for a long time they were almost 100% compatible at the driver
level (something you could never really say about net and free).
Warner
On Mon, Jan 20, 2020, 10:21 AM Clem Cole <clemc(a)ccc.com> wrote:
On Sun, Jan 19, 2020 at 10:52 PM George Michaelson <ggm(a)algebras.org>
wrote:
It does me no credit, that I initially reacted
very badly to 386BSD,
and the initial {Net,Free,Open} situation.
First, be careful. What we sometimes call 386BSD as a 'release'
started
just as a port of NET2 to the 386 based 'commodity' hardware platform. The
history is that in the late 1970s/early 80s Bill Jolitz was working for Nat
Semi and ported BSD 4.1, to a multibus based NS16032 board that NS had
built, which was similar to the Stanford University Network (SUN) terminal
what had a 68000. He eventually built a 'luggable' using that and
updated to the port to 4.2++. He (and Lynn I believe) started a company
to sell that hardware/software solution and for whatever reason, it did not
really take off.
At some point, he got his hands on a 386based PC (Compaq I think) and
started to move his port over to that system. A number of people helped
him (for instance I did a bunch of the AT/disk controller work as I had
access to the WD design documents for another consulting gig I had at the
time - Bill mentioned this in the articles BTW).
Bill and Lynn's NS16032 and 386 code went back to the CSRG 'masters' -
although how and that happened was never completely clear to me. The SCCS
deltas tell at least part of the story. Bill managed to make a bootable
image that mostly installed on a PC/386 as the minicomputer versions did
from the formal release. The ftp area of ucbvax had all of these bootable
images available for download such as one for an HP 68K system and I think
the DEC VAX and PMAX, the CCG system and a few others IIRC. As I have said
in other messages if you were a UCB licensee you had the passwords to
look/download from that area. Bill placed that version in the same ftp
area. The 386 based port went viral at least with the UCB licensees. (In
fact, if Linus had known about it, theoretically he could have used it
also. His university was licensee, but as Larry McVoy likes to point, not
all the schools were as free with the IP, so I will not go down that
rathole).
The bottom line is that many people (like me on a Wyse386) started with
Bill's original port; including the BSDi founders.
When Jolitz and BSDi went separate ways, Jolitiz continued to update the
CSRG 386 based tarball (to an extent). One of the issues was there
originally was attempt to keep the different architectural versions of BSD
in sync ( to a point and NetBSD does yet exist). A number of people were
unhappy and the speed, depth *etc*. of the 386 version, most
notably Jordan Hubbard and FreeBSD was born. The two biggest issues Jordan
wanted to fix, was easier install and a bit wider support for more hardware
(again I sent Jordan the changes to FreeBSN 1x for the Wyse and a couple of
NCR boxes). The NetBSD project would birth from the original ideals of
CSRG and trying to keep everything the same but that's still in the future.
I found all this "fragmentation" pretty hard to understand. -BSDI felt
like it had occupied the space, and I couldn't entirely understand
what was going on, or why any of it mattered.
See below....
What I think I missed (didn't understand)
was how draining support was
for Berkeley, and in the absence of a sugar daddy
Herein is the issue that many people on the sidelines missed.
CRSG was a large project and funded a lot of work at UCB in EECS. It never
funded me (I was funded by Tektronix, HP, DEC *et al*), but that project
did a fund a number of students. However, at some point CSRG stopped being
a research project and started being a support project for DARPA. There
was also a good deal of resentment by some groups in EECS that were not
getting DARPA funding.
I'll not say if that was good or bad but I will say that it did cause
great deal consternation at UCB within the department and many people doing
more formal research were not happy. In the end, the EECS
Department mothers and fathers along with the Dean *et al*, decided to
stop/end the CSRG project. Many people who were directly or indirectly
working on BSD, like Mary Ann and myself, had graduated and had since
left. Bob Kridle had formed Mt. Xinu, Asa Romberger has formed Unisoft, Joy
had left/was leaving for Sun, *etc*. So the question remained what to
do with CSRG. As to what everyone would do, became every person for
her/himself and as we know some of the folks, along with a few folks from
the USENIX community formed BSDi.
As was noted elsewhere, NetBSD would eventually be formed by volunteers to
keep the different ports alive (in fact much of the efforts was from folks
not at UCB), but that was still in the offing. Remember, while CSRG
itself was not a research project, a lot of people around the world were
using the BSD code base for their own research. The whole idea of NetBSD
was to create a uniform platform that people could compare things. So, the
question of how that was to come about or do any work on BSD if DARPA was
not paying the bills, was still an open one. But, the idea that would
eventually create FreeBSD, was supporting a pure commodity *solution for
day-to-day use, not as a research platform*. [I'll leave off the later
OpenBSD/NetBSD fork by Theo here as it has little to do with the question].
BSDi had a similar/same goal of producing something like SunOS/VMS *etc* but
supported on commodity hardware. That solution was to sell it and using
the revenues from the support contract, be able to pay people to do that
work. As I said and in some other messages, it is noted that Bill Jolitz
wanted something more FOSS. Truth is BSDi code was 'open source' but it
took a $1K license to *get the source from them*.
In the end, the real problem was not the infighting between the different
BSD camps, but AT&T, who wanted the entire pie. Clearly, their executives
saw anything other than their complete control of the UNIX IP as a threat.
Hence the court case, the eventually AT&T/Sun relationship *etc*...
Your lack of 'sugar daddy,' really comes back to that. There were few
people at the time that could pay the bills. Until then DARPA had been
it. I do not know if DARPA wanted out or if another group could have been
formed that could take over CSRG. I did have discussions with Rob over a
beer that at least the thought had crossed the BSDi folks mind, that once
started; they would apply for a DARPA contract.
At the time had blow up, I was a consultant and I personally was
considering what I was going to do next and if they had had a real future,
the talks with Rob might have gotten more serious. My wife wanted me to
stop being independent if we were to start a family (I would join Locus
instead).
BTW: I was in an interesting position as I was friends with all of the
different sides in the war/original fight. Like Jolitz, I wanted to see
what we now call a 'FOSS' release of BSD. But like Rob, I knew it was
going to take some revenue stream to make it happen/continue the support.
In the end, the AT&T legal mess blew it all up. BSDi ended up failing
and Jordan's work stayed around.
BTW: what pays for Linux development these days by number of 'committers
salary' is Intel (#1), IBM (#2), then a load of other firms including the
different distros. But for *any* platform to be successful and actually
continue to be used in the market, someone has to pay the salaries of some
set of professional programmers to do the work.
That said when AT&T injoined BSDi and UCB a lot of people (myself
included) started to hack on Linux. But just think if AT&T had actually
won the case and courts decided UNIX was allowed to be a trade secret, then
Linux and all of the UNIX 'clones' would have been in violation.
No matter what flavor of UNIX you like, we are all in debt to UCB and BSDi
for settling the IP argument. The court was clear, the >>ideas<< behind
UNIX (*a.k.a.* the intellectual property) came from Ken, Dennis and
friends at AT&T and *they did own it.* But because of the 1956 consent
decree that published the ideas and the moment the ideas were published, we
all can now >>use<< them. The provenance of the source code does not
relate to the provenance of the idea, so* the source code itself does not
define what UNIX is or is not. *
I bring this all up in hopes to try to close this rat hole of Linux, *vs*.
*BSD. Like editors, we all have our own favorites. That's cool, we don't
want one thing to be forced down our throat. Having a choice is what is
good. And what I value, Larry or Jon may not necessarily like. Most of
us if not all on this list probably want something that approximates Ken
and Dennis's original ideas not what IBM, DEC, CDC were trying to make us
use in the old days or what Microsoft calls a system today.
The discussion of how we got there and what people valued at the time is
useful so we can try to remember the history and learn from it; but getting
into right/wrong, good/bad, or you could have had this is a tad tiresome;
IMO.
7 p.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
On Mon, Jan 20, 2020 at 12:49 PM Warner Losh <imp(a)bsdimp.com> wrote:
Thanks Clem.
Most welcome
One minor clarification. Jordan and the patchkit work did predate NetBSD.
However, the NetBSD project formed a little before the FreeBSD project that
grew out of the patchkit days. Jordan didn't get that moving until NetBSD
made rumblings...
Right - if I was not clear, on that ordering, mei culpa.
it was still a time that you heard a lot of what was
going on by word of
mouth, not so much by postings and email...
Exactly.
The OpenBSD split was years later... and a complicated mix of personality
conflicts and technical differences.
A real shame IMO, but giving the personalities, I'm not sure it was not
predestined,
But in many ways it was a smaller split since for a
long time they were
almost 100% compatible at the driver level
Very true, I run OpenBSD on my router/main server - I just want a minimum
system, that I feel it safe.
I have NetBSD on a couple of boxes cause it runs and FreeBSD or Linux on
others. And MacOS on my desktop.
I'm sort of, whatever gets the job done and I don't have to think too much
about it, but it's probably why the little incompatibilities drive me nuts.
6:04 p.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
On Mon, Jan 20, 2020 at 12:19:25PM -0500, Clem Cole wrote:
On Sun, Jan 19, 2020 at 10:52 PM George Michaelson
<ggm(a)algebras.org> wrote:
I know those Nat Semi chips very well, or did at the time. I so wanted to
love those chips, the instruction set felt like whoever did the PDP-11
did the 320xx chips. But they couldn't produce chips without bugs and
that killed them. It's a crying shame, I liked the instruction set
WAY better than the VAX. The VAX seemed really messing compared to
the PDP-11, the 320xx chips seemed clean. Might be rose colored
glasses but that's my memory.
It does me no credit, that I initially reacted
very badly to 386BSD,
and the initial {Net,Free,Open} situation.
First, be careful. What we sometimes call 386BSD as a 'release'
started
just as a port of NET2 to the 386 based 'commodity' hardware platform. The
history is that in the late 1970s/early 80s Bill Jolitz was working for Nat
Semi and ported BSD 4.1, to a multibus based NS16032 board that NS had
built, which was similar to the Stanford University Network (SUN) terminal
what had a 68000. He eventually built a 'luggable' using that and updated
to the port to 4.2++. He (and Lynn I believe) started a company to sell
that hardware/software solution and for whatever reason, it did not really
take off. 
6:09 p.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
On Jan 20, 2020, at 10:04 AM, Larry McVoy <lm(a)mcvoy.com> wrote:
On Mon, Jan 20, 2020 at 12:19:25PM -0500, Clem Cole wrote:
My memory as well. A friend and I got ahold of the complete set of chips
and started to build out the hardware for a Unix box. We got most of the
way there too, and then the odd quirks started showing up. We tracked
some of them to our layout and the others to the NS chips. Then we gave
it up as a “ah, it would have been nice if only” project.
David
On Sun, Jan 19, 2020 at 10:52 PM George
Michaelson <ggm(a)algebras.org> wrote:
I know those Nat Semi chips very well, or did at the time. I so wanted to
love those chips, the instruction set felt like whoever did the PDP-11
did the 320xx chips. But they couldn't produce chips without bugs and
that killed them. It's a crying shame, I liked the instruction set
WAY better than the VAX. The VAX seemed really messing compared to
the PDP-11, the 320xx chips seemed clean. Might be rose colored
glasses but that's my memory. It does me no credit, that I initially reacted
very badly to 386BSD,
and the initial {Net,Free,Open} situation.
First, be careful. What we sometimes call 386BSD as a 'release'
started
just as a port of NET2 to the 386 based 'commodity' hardware platform. The
history is that in the late 1970s/early 80s Bill Jolitz was working for Nat
Semi and ported BSD 4.1, to a multibus based NS16032 board that NS had
built, which was similar to the Stanford University Network (SUN) terminal
what had a 68000. He eventually built a 'luggable' using that and updated
to the port to 4.2++. He (and Lynn I believe) started a company to sell
that hardware/software solution and for whatever reason, it did not really
take off. 
6:34 p.m.
New subject: Early Linux and BSD
On 1/20/2020 1:09 PM, David Barto wrote:
My memory as well. A friend and I got ahold of the
complete set of chips
and started to build out the hardware for a Unix box. We got most of the
way there too, and then the odd quirks started showing up. We tracked
some of them to our layout and the others to the NS chips. Then we gave
it up as a “ah, it would have been nice if only” project.
A friend of mine and I
did the same, but with 68000. I wrote the
assembler and other utilities, he did the hardware design (although we
both took turns critiquing each other's work), and we both (and his
girlfriend) did the wire-wrapping.
We were eventually consumed by other real-life happenings, and it fell
by the wayside, but looking back at it now, UNIX would have been the
perfect choice. I did have visions of grandeur that I would write my own
OS for it - I'm sure I would have eventually learned my lesson ;)
art k.
PS: We cheated, and used static RAM. A few failed development projects
at my (then) current place of employment, and no one knew what to do
with "all these chips" that were left over. No problem, I know what to
do with them... they wound up in the trunk of my Triumph TR7
7:18 p.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
On Mon, Jan 20, 2020 at 1:04 PM Larry McVoy <lm(a)mcvoy.com> wrote:
I know those Nat Semi chips very well, or did at the
time. I so wanted to
love those chips, the instruction set felt like whoever did the PDP-11
did the 320xx chips.
It was clear, the NS folks, like the Moto folk, knew the PDP-11 and VAX.
It was a nice architecture and should been a win but...
But they couldn't produce chips without bugs and
that killed them.
I really think it was they were the third man and too late. Between Apple
on the Mac and Apollo, Masscomp and eventually Sun, the 68000 and later
68010 had volume. 8086 family had volume with the PC.
As Jon can tell you, Tektronix decided the use the NS chip and tossed a
working 68000/68010 design (Magnolia - which would later ship at 4400) for
a completely new workstation. But it meant starting from scratch. Big
mistake...
If they had just shipped Magnolia at the beginning, I'm not sure either
Masscomp or Sun would have been birthed. Apollo and Triple-Drip were
already there, but thy would have had the first UNIX workstation on the
market, with a super graphics system. Sigh....
It's a crying shame, I liked the instruction set
WAY better than the VAX. The VAX seemed really messing compared to
the PDP-11, the 320xx chips seemed clean. Might be rose colored
glasses but that's my memory.
I would not say way better, but much cleaner. To DEC's credit, the idea of
the VAX was to take the PDP-11 forward and keep things running. Funny, DG
did that better in the end, but that was the idea at least. NS was trying
to make the VAX without the rough edges, mistakes DEC had made. Prof. Yale
Patt consulted on both Instructions sets BTW, which may be why they look so
similar.
7:46 p.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
Clem Cole writes:
On Mon, Jan 20, 2020 at 1:04 PM Larry McVoy <lm(a)mcvoy.com> wrote:
I remember it slightly differently than Clem, but close. The Magnolia wasn't
a UNIX workstation, it was an experimental Smalltalk machine. I don't recall
much about it, but I don't think that it had to address many of the problems
that UNIX had at the time with the 68000 such as the lack of a MMU. I think
that the Magnolia predated the 68010 and certainly predated the 68020 and
awful but usable PMMU. There were also many political issues because by this
time the legacy of Howard Vollum had departed Tektronix and it was starting to
die the slow death of a poorly managed company being looted by top management
which has become all too common since. But at least Tek lead in something!
Part of the issue was that the Magnolia was developed in Tek Labs, which was
the research end of things. It wasn't a product organization, the Magnolia
at the time hadn't gone through any of the rigorous environmental testing
required by Tek which was a company that actually provided warranty service.
And there was no marketing, not that Tek was a marketing powerhouse. Given
the way that things panned out I don't think that the Magnolia would have been
a player once things like Suns appeared, if for no other reason that Tek had no
clue as to how to do anything in volume and our stuff was way too expensive.
The 32032 made sense for the workstation division based on the data sheets.
But, it turned out to be extremely buggy, and unlike the 68K I don't recall
the ability to look at and patch the state of the microcode.
In any case, while the 32032 was a problem, the real reason that Tek failed
in the workstation biz was management. What happened was that programmable
instrumentation was becoming a thing. Every instrument group in the company
was building their own "controller" for instrument programming. In a rare
case of good decision-making it was decided that a single group would build
a controller that everyone else would use; this was the workstation division.
But, this took the most fun thing away from all of the other groups. They
way that management structured things, instrument groups have to use the
workstation unless it was missing something that they required. The result
was that the workstation had to meet the union of all requirements, real or
imagined. The I/O board in this thing had like 4 GPIB ports, 24 RS-232 ports,
8 RS-422 ports (I don't remember the exact number), and so on, making it more
expensive than anybody else's CPU board. Of course, when the IBM PC came along
all of the instrument groups said "well, we have 2 RS-232 ports and a parallel
port and so we'll work with that." Had management said that the workstation
group could do what was reasonable and that everyone would have to adapt and
use it we could have parallel groups competing on 32K and 68K designs.
Jon
I know those Nat Semi chips very well, or did at
the time. I so wanted to
love those chips, the instruction set felt like whoever did the PDP-11
did the 320xx chips.
It was clear, the NS folks, like the Moto folk, knew the PDP-11 and VAX.
It was a nice architecture and should been a win but...
But they couldn't produce chips without bugs
and that killed them.
I really think it was they were the third man and too late. Between Apple
on the Mac and Apollo, Masscomp and eventually Sun, the 68000 and later
68010 had volume. 8086 family had volume with the PC.
As Jon can tell you, Tektronix decided the use the NS chip and tossed a
working 68000/68010 design (Magnolia - which would later ship at 4400) for
a completely new workstation. But it meant starting from scratch. Big
mistake...
If they had just shipped Magnolia at the beginning, I'm not sure either
Masscomp or Sun would have been birthed. Apollo and Triple-Drip were
already there, but thy would have had the first UNIX workstation on the
market, with a super graphics system. Sigh....
It's a crying shame, I liked the instruction
set
WAY better than the VAX. The VAX seemed really messing compared to
the PDP-11, the 320xx chips seemed clean. Might be rose colored
glasses but that's my memory.
I would not say way better, but much cleaner. To DEC's credit, the idea of
the VAX was to take the PDP-11 forward and keep things running. Funny, DG
did that better in the end, but that was the idea at least. NS was trying
to make the VAX without the rough edges, mistakes DEC had made. Prof. Yale
Patt consulted on both Instructions sets BTW, which may be why they look so
similar.
8:15 p.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
On Mon, Jan 20, 2020 at 2:47 PM Jon Steinhart <jon(a)fourwinds.com> wrote:
I remember it slightly differently than Clem, but
close.
Ouch -- I was 1/2 of the Magnolia development team -- I remember a lot
about it!!!
For the curious when bitsavers comes back: <goog_573452328>
http://bitsavers.trailing-edge.com/pdf/tektronix/magnolia/
Roger Bates had just finished the Dorado at PARC. I had just left CMU.
We were cubical mates in TekLabs. Motorola had an experimental chip that
was not yet numbered. We were given them in the Computer Research group
in Tek Labs. So, we started building a personal computer at night for
ourselves.
Our boss saw the notes and asked what we would do differently if Tek paid
for it. I was originally using 8" floppies and immediately said 'a real
disk.' We got a Tek '$10K project' and a few months to build a
prototype.
I already had written (well sort of hacked) a simple C compiler based on
Dennis's PDP-11 compiler (when it screwed up it would sometimes include
PDP-11 code - and I never supported FP). Paul Blattner wrote an assembler
and linker. Using that, Steve Glaser and I ported UNIX/V7 to it.
The Magnolia wasn't a UNIX workstation, it was an
experimental Smalltalk
machine.
That was 2+ years later actually. Once they had the system, a couple of
other folks moved Smalltalk to it. And in fact, it eventually did release
it as a product called the 4404.
I don't recall
much about it, but I don't think that it had to address many of the
problems
that UNIX had at the time with the 68000 such as the lack of a MMU.
Be careful... It most definitely did have an MMU, I designed it!!! The
Xerox Altos and Dorado's never had MMU's. So Roger was not familiar with
them. I had to teach him. Magnolia had a base/limit register MMU similar
to the PDP-11/70. The original OS was V7 and swapped. It ran just fine.
I think
that the Magnolia predated the 68010 and certainly predated the 68020 and
awful but usable PMMU.
The wire-wrapped prototype was originally an X-series chip and yes the
first 'production' units were real 10Mhz 68000s. After I went back to
grad school, Roger spliced a 68010 into and ripped out my MMU. The late
Terry Laskodi put 4.1BSD on it.
Part of the issue was that the Magnolia was developed
in Tek Labs, which
was
the research end of things. It wasn't a product organization, the Magnolia
at the time hadn't gone through any of the rigorous environmental testing
required by Tek which was a company that actually provided warranty
service.
And there was no marketing, not that Tek was a marketing powerhouse. Given
the way that things panned out I don't think that the Magnolia would have
been
a player once things like Suns appeared, if for no other reason that Tek
had no
clue as to how to do anything in volume and our stuff was way too
expensive.
Very possible, but they did have first mover position. In fact, folks at
Harvard Business as much as said so later. There is a great HBS case study
written about it called "Why Skunk Projects Don't Work" (which I have
somewhere) -- I should get that scanned at added to the Magnolia archive on
BitSavers.
In any case, while the 32032 was a problem, the real reason that Tek failed
in the workstation biz was management.
No doubt... but it was 3 years later. Which I think was a huge issue.
"well, we have 2 RS-232 ports and a parallel
port and so we'll work with
that."
Which of course was what Magnolia had been 3.5 years earlier and was what
became the 4404 Smalltalk machine.
21 Jan
21 Jan
6:58 a.m.
New subject: Early Linux and BSD
Jon Steinhart wrote:
The 32032 made sense for the workstation division
based on the data sheets.
But, it turned out to be extremely buggy, and unlike the 68K I don't recall
the ability to look at and patch the state of the microcode.
Did you have the ability to look at and patch the state of 68000
microcode? How?
2:30 p.m.
New subject: Early Linux and BSD
It's possible they added an internal register or two to help save the fault
State but I don't think so. We'd have to ask Les Crudele or Nick T. to be
sure. As I understand it from my days of commuting with Les to Stellar
and his talking about the project on those trips, the only real difference
between the 68000 and 68010 was Nick's microcode which was in the mask. I
think they may have stepped the mask in a few places for better yields but
I understand the impression that was not externally electrical different in
function.
[Les was the 68000's main logic designer and Nick wrote the microcode].
On Tue, Jan 21, 2020 at 1:59 AM Lars Brinkhoff <lars(a)nocrew.org> wrote:
Jon Steinhart wrote:
--
Sent from a handheld expect more typos than usual
The 32032 made sense for the workstation division
based on the data
sheets.
But, it turned out to be extremely buggy, and
unlike the 68K I don't
recall
the ability to look at and patch the state of the
microcode.
Did you have the ability to look at and patch the state of 68000
microcode? How?
5:17 p.m.
New subject: Early Linux and BSD
Lars Brinkhoff writes:
Jon Steinhart wrote:
My memory is very very very fuzzy on this. I seem to recall that microcode
state was pushed onto a stack in certain cases, and that it was possible to
fix some problems there for certain weird cases relating to memory management.
That's all that I remember about it as that's not the part of things that I
was working on, just heard grumbles from other folks about it.
Jon
The 32032 made sense for the workstation division
based on the data sheets.
But, it turned out to be extremely buggy, and unlike the 68K I don't recall
the ability to look at and patch the state of the microcode.
Did you have the ability to look at and patch the state of 68000
microcode? How?
5:22 p.m.
New subject: Early Linux and BSD
On Tue, Jan 21, 2020, 10:18 AM Jon Steinhart <jon(a)fourwinds.com> wrote:
Lars Brinkhoff writes:
My memory is very very very fuzzy on this. I seem to recall that microcode
state was pushed onto a stack in certain cases, and that it was possible to
fix some problems there for certain weird cases relating to memory
management.
That's all that I remember about it as that's not the part of things that I
was working on, just heard grumbles from other folks about it.
This isn't for the two cpu design to allow instructions to be restarted
after a page fault.
Warner
Jon
Jon Steinhart wrote:
> The 32032 made sense for the workstation division based on the data
sheets.
> But, it turned out to be extremely buggy,
and unlike the 68K I don't
recall
the
ability to look at and patch the state of the microcode.
Did you have the ability to look at and patch the state of 68000
microcode? How?
5:25 p.m.
New subject: Early Linux and BSD
Warner Losh writes:
On Tue, Jan 21, 2020, 10:18 AM Jon Steinhart <jon(a)fourwinds.com> wrote:
No, this was using a 68020 with the PMMU.
Lars Brinkhoff writes:
My memory is very very very fuzzy on this. I seem to recall that microcode
state was pushed onto a stack in certain cases, and that it was possible to
fix some problems there for certain weird cases relating to memory
management.
That's all that I remember about it as that's not the part of things that I
was working on, just heard grumbles from other folks about it.
This isn't for the two cpu design to allow instructions to be restarted
after a page fault.
Warner
Jon Jon Steinhart wrote:
> The 32032 made sense for the workstation division based on the data
sheets.
> But, it turned out to be extremely buggy,
and unlike the 68K I don't
recall
the
ability to look at and patch the state of the microcode.
Did you have the ability to look at and patch the state of 68000
microcode? How?
6:43 p.m.
New subject: Early Linux and BSD
On Tue, Jan 21, 2020 at 12:18 PM Jon Steinhart <jon(a)fourwinds.com> wrote:
My memory is very very very fuzzy on this. I seem to
recall that microcode
state was pushed onto a stack in certain cases,
State, not the code.
In fact, Masscomp having built the first MP UNIX box, ran into this problem
early on. Different processor stepping had different internal microcode
state on the stack after an IRQ. If you resumed with a processor that was
a different processor revision, the wrong state was returned.
Will may remember this, but Masscomp issues strick orders to the FE that
all CPU boards had to be the revision. You could not just swap a CPU
board, they had to go as sets. It was a real bummer.
Moto fixed that with the 020 and later devices as more people made MP
systems.
... just heard grumbles from other folks about it.
Probably me ... it took me, tjt and Terry Hayes about 3-4 weeks to figure
out that problem. It was not originally documented, other than to state
on certain faults X bytes of reserved information was pushed on the stack.
BTS: I don't remember, but it may have started with the 68010.
Becuase before that, the 'executor' was wait stated and the fixor handled
and fixed the fault so the 68000 never actually saw fault in the original
Masscomp CPU board. The "MPU" board was the same board with a couple of
PAL's changed and an 68010 as the executor. It was allowed to actually
fault and do something else while the fixor corrected the fault. But the
key is that when the fault was repaired, another executor on a different
MPU board could be the processor that 'returned' from the fault. That
ended up being a no-no.
6:44 p.m.
New subject: Early Linux and BSD
sorry... all *MPU* boards had to be the revision but we may have done
the same with the CPU boards.
On Tue, Jan 21, 2020 at 1:43 PM Clem Cole <clemc(a)ccc.com> wrote:
On Tue, Jan 21, 2020 at 12:18 PM Jon Steinhart <jon(a)fourwinds.com> wrote:
My memory is very very very fuzzy on this. I
seem to recall that
microcode
state was pushed onto a stack in certain cases,
State, not the code.
In fact, Masscomp having built the first MP UNIX box, ran into this
problem early on. Different processor stepping had different internal
microcode state on the stack after an IRQ. If you resumed with a processor
that was a different processor revision, the wrong state was returned.
Will may remember this, but Masscomp issues strick orders to the FE that
all CPU boards had to be the revision. You could not just swap a CPU
board, they had to go as sets. It was a real bummer.
Moto fixed that with the 020 and later devices as more people made MP
systems.
... just heard grumbles from other folks about
it.
Probably me ... it took me, tjt and Terry Hayes about 3-4 weeks to figure
out that problem. It was not originally documented, other than to state
on certain faults X bytes of reserved information was pushed on the stack.
BTS: I don't remember, but it may have started with the 68010.
Becuase before that, the 'executor' was wait stated and the fixor handled
and fixed the fault so the 68000 never actually saw fault in the original
Masscomp CPU board. The "MPU" board was the same board with a couple of
PAL's changed and an 68010 as the executor. It was allowed to actually
fault and do something else while the fixor corrected the fault. But the
key is that when the fault was repaired, another executor on a different
MPU board could be the processor that 'returned' from the fault. That
ended up being a no-no.
7:14 p.m.
New subject: Early Linux and BSD
On Tue, Jan 21, 2020, 11:46 AM Clem Cole <clemc(a)ccc.com> wrote:
sorry... all *MPU* boards had to be the revision
but we may have done
the same with the CPU boards.
When did Masscomp ship their first MP system?
Warner
On Tue, Jan 21, 2020 at 1:43 PM Clem Cole <clemc(a)ccc.com> wrote:
On Tue, Jan 21, 2020 at 12:18 PM Jon Steinhart <jon(a)fourwinds.com> wrote:
My memory is very very very fuzzy on this. I
seem to recall that
microcode
state was pushed onto a stack in certain cases,
State, not the code.
In fact, Masscomp having built the first MP UNIX box, ran into this
problem early on. Different processor stepping had different internal
microcode state on the stack after an IRQ. If you resumed with a processor
that was a different processor revision, the wrong state was returned.
Will may remember this, but Masscomp issues strick orders to the FE that
all CPU boards had to be the revision. You could not just swap a CPU
board, they had to go as sets. It was a real bummer.
Moto fixed that with the 020 and later devices as more people made MP
systems.
... just heard grumbles from other folks about
it.
Probably me ... it took me, tjt and Terry Hayes about 3-4 weeks to
figure out that problem. It was not originally documented, other than
to state on certain faults X bytes of reserved information was pushed on
the stack.
BTS: I don't remember, but it may have started with the 68010.
Becuase before that, the 'executor' was wait stated and the fixor handled
and fixed the fault so the 68000 never actually saw fault in the original
Masscomp CPU board. The "MPU" board was the same board with a couple of
PAL's changed and an 68010 as the executor. It was allowed to actually
fault and do something else while the fixor corrected the fault. But the
key is that when the fault was repaired, another executor on a different
MPU board could be the processor that 'returned' from the fault. That
ended up being a no-no.
8:27 p.m.
New subject: Early Linux and BSD
1982, the dual-processor MC500/DP originally with 68000s upgraded to 010's
shortly after they became available[see below]
1984, the 16 Processor MC5000/700 using '020 [the 500 was renamed the
MC5000/500 and a single processor MC5000/300 was also introduced. In the
/700 and /300 design the fixor was unneeded and the base 020 serviced it's
own faults].
FWIW: Purdue VAX predates the 500/DP, but was a one-off that George made.
The Sequent MP box would be about 3 or 4 years later.
Through the RTU 2.x, the OS originally ran Purdue VAX-like [*Goble/Marsh: ISCA
'82: Proceedings of the 9th annual symposium on Computer Architecture: "A
Dual Processor VAX 11/780", **Pages 291–298*] in all interrupts and system
calls went to a 'master' and the second MPU/CPU board ran as a
'slave' (
*i.e.* user-mode code). By RTU 3.0 ~12 mons later, full locks were done and
each processor could service anything.
Note each CPU/MPU board had processor two chips on it, the executor and
fixor but the board was really not a multiprocessor - the second chip was
literally just running kernel code to service the page fault. Thus (not
including the other 68000's processors in graphics or I/O boards) the
500/DP had either 4 68000's or 2 68010 & 2 68000's in it when it had two
CPU or two MPU boards in the backplane. The idea was originally proposed
for the Z8000 by Forest Baskett at an early Asilomar conference. The
formal citation is: Forest Baskett: "*Pascal and Virtual Memory in a Z8000
or MC68000 based Design Station*," COMPCON 80, Digest of Papers, pp
456-459, IEEE Computer Society, Feb. 25, 1980.
On Tue, Jan 21, 2020 at 2:14 PM Warner Losh <imp(a)bsdimp.com> wrote:
On Tue, Jan 21, 2020, 11:46 AM Clem Cole <clemc(a)ccc.com> wrote:
sorry... all *MPU* boards had to be the
revision but we may have done
the same with the CPU boards.
When did Masscomp ship their first MP system?
Warner
On Tue, Jan 21, 2020 at 1:43 PM Clem Cole <clemc(a)ccc.com> wrote:
>
>>
>>
>> On Tue, Jan 21, 2020 at 12:18 PM Jon Steinhart <jon(a)fourwinds.com>
>> wrote:
>>
>>> My memory is very very very fuzzy on this. I seem to recall that
>>> microcode
>>> state was pushed onto a stack in certain cases,
>>
>> State, not the code.
>>
>> In fact, Masscomp having built the first MP UNIX box, ran into this
>> problem early on. Different processor stepping had different internal
>> microcode state on the stack after an IRQ. If you resumed with a processor
>> that was a different processor revision, the wrong state was returned.
>>
>> Will may remember this, but Masscomp issues strick orders to the FE that
>> all CPU boards had to be the revision. You could not just swap a CPU
>> board, they had to go as sets. It was a real bummer.
>>
>> Moto fixed that with the 020 and later devices as more people made MP
>> systems.
>>
>>
>>
>>
>>
>>> ... just heard grumbles from other folks about it.
>>>
>> Probably me ... it took me, tjt and Terry Hayes about 3-4 weeks to
>> figure out that problem. It was not originally documented, other than
>> to state on certain faults X bytes of reserved information was pushed on
>> the stack.
>>
>> BTS: I don't remember, but it may have started with the 68010.
>> Becuase before that, the 'executor' was wait stated and the fixor
handled
>> and fixed the fault so the 68000 never actually saw fault in the original
>> Masscomp CPU board. The "MPU" board was the same board with a couple
of
>> PAL's changed and an 68010 as the executor. It was allowed to actually
>> fault and do something else while the fixor corrected the fault. But the
>> key is that when the fault was repaired, another executor on a different
>> MPU board could be the processor that 'returned' from the fault.
That
>> ended up being a no-no.
>>
>
22 Jan
22 Jan
12:14 a.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
At Mon, 20 Jan 2020 10:04:32 -0800, Larry McVoy <lm(a)mcvoy.com> wrote:
Subject: Re: [TUHS] Early Linux and BSD (was: On the origins of Linux - "an academic
question")
I know those Nat Semi chips very well, or did at the time. I so wanted to
love those chips, the instruction set felt like whoever did the PDP-11
did the 320xx chips. But they couldn't produce chips without bugs and
that killed them. It's a crying shame, I liked the instruction set
WAY better than the VAX. The VAX seemed really messing compared to
the PDP-11, the 320xx chips seemed clean. Might be rose colored
glasses but that's my memory.
I held a lot of anticipation for the NS chips as well, and I remember
well excitedly going around to trade shows for a year or two and playing
around with the very few Unix systems based on them that showed up on
occasion.
From what I understand it was really only the original NS32016 that was
too buggy to be trusted. The NS32032 was a bug-fix release that also
came a full 32-bit external bus, and the NS32532 a while later was quite
a contender at the time in terms of performance (wikipedia says "about
twice as performant as the competing MC68030 and i80386").
In the end though I discovered the ATT 3b2 systems and their also quite
nicely orthogonal WE32000 CPUs (though in the end I never did write more
than a very simple demo program in assembler, just to know I could). My
copy of the WE32100 Information Manual sits right beside my VAX
Instruction Reference Manual. Sadly the 3B2s I had were never as
powerful as a PC532 was though -- more like the sluggish i386 and m68k
systems of the day.
--
Greg A. Woods <gwoods(a)acm.org>
Kelowna, BC +1 250 762-7675 RoboHack <woods(a)robohack.ca>
Planix, Inc. <woods(a)planix.com> Avoncote Farms <woods(a)avoncote.ca>
21 Jan
21 Jan
12:44 a.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
On Mon, 20 Jan 2020 12:19:25 -0500 Clem Cole <clemc(a)ccc.com> wrote:
I bring this all up in hopes to try to close this rat hole of Linux, *vs*.
Ha!
*BSD. Like editors, we all have our own favorites.
That's cool, we don't
want one thing to be forced down our throat. Having a choice is what is
good. And what I value, Larry or Jon may not necessarily like. Most of
us if not all on this list probably want something that approximates Ken
and Dennis's original ideas not what IBM, DEC, CDC were trying to make us
use in the old days or what Microsoft calls a system today.
I want an *evolution* of their original ideas. Instead we have
significantly more complicated systems which may be more
efficient but certainly a lot more bloaty and less flexible
(with a few exceptions). [Ties in with the single system idea]
The discussion of how we got there and what people
valued at the time is
useful so we can try to remember the history and learn from it; but getting
into right/wrong, good/bad, or you could have had this is a tad tiresome;
IMO.
+1
20 Jan
20 Jan
7:09 p.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
On Sun, Jan 19, 2020 at 07:32:57PM -0800, Greg A. Woods wrote:
.... which is dated July 1992, and describes a "launch" of 386BSD
Release 0.0 in March 17, 1992. This is contemporaneous with Linux
0.95a (which by coincidence was also released on March 17th, 1992.)
The first "real" distribution, the Soft Landing System, was released
in May 1992. (The Manchester Computer Centre distribution in November
1991 was a floppy-based distro containing command-line and development
utilities, but not X Windows, so some people don't feel it counts as a
full-featured distribution.)
Out of
curiosity, did the articles contain download information for a
bootable copy of 386BSD?
Yes, they did:
https://www.drdobbs.com/porting-unix-to-the-386-the-final-step/184408800 Also keep in mind that NetBSD started as a set of
"net" (as in usenet)
patch kits for 386BSD.
It looks like NetBSD's source code repository was established on March
21, 1993. Patchkit 0.2.2 was apparently also released on the same
date. NetBSD's first release, 0.8, was released on April 19, 1993.
The FreeBSD project was named in June 19, 1993, with its first release
in November 1993.
So it's easy to use the lawsuit as the scapegoat for why the BSD's
failed to take off, but at best it's only one of many factors. The
Jolitzs' refusal to accept many patches, forcing a delay of a year
before spawning two project forks, was one. The dispersal of effort
as a side effect of various people trying to start companies around
BSD code (SunOS, NetApp, BSDI, Wasabi Systems, etc.) was another.
BSD-licensed code seems to thrive best when there are grants or
non-profit institutions funding its work; but attempts to support BSD
code from as part of commercial work doesn't seem to have worked out
as well.
As dwheeler (I think Dave Wheeler, but I'm not certain) astutely
observed in 2006:
I think the BSD license has been a lot of trouble to the
*BSDs. Every few years, someone says, "hey, let's start a company
based on this BSD code!" (BSD/OS in particular comes to mind, but
SunOS and others did the same). They pull the *BSD code in, and
some of the best BSD developers, and write a proprietary
derivative. But as a proprietary vendor, their fork becomes
expensive to self-maintain, and eventually the company
founders. All that company work is lost forever, and good
developers were sucked away during that period. Repeat, repeat,
repeat. That's more than enough to explain why the BSDs manage to
make steps forward, but just don't manage to maintain the pace of
Linux kernel development.
Meanwhile, the GPL has legally enforced a consortia on major
commercial companies. Red Hat, Novell, IBM, and many others are all
contributing, and feel safe in doing so because the others are
legally required to do the same. It's basically created a "safe"
zone of cooperation, without anyone having to sign complicated
legal documents. A company can't feel safe contributing code to the
BSDs, because its competitors might simply copy it without
reciprocating. There's much more corporate cooperation in the
GPL'ed kernel code than with the BSD'd kernel code. Which means
that in practice, it's actually been the GPL that's most
"business-friendly". So while the BSDs have lost energy every time
a company gets involved, the GPL'ed programs gain almost every time
a company gets involved. And that explains it all.
- https://lwn.net/Articles/197875/
I'll also note that the GPL licensing means that I've been able to
carry my expertise in the code base across 4 job changes (MIT, VA
Linux Systems, IBM, Google). In effect, this arrangement and the
business models forced by the GPL allocates more value to the
community at large and to the engineers working at those companies, at
the expense of value that can be extracted to the corporate
shareholders --- for better or for worse.
And so while I don't have a private jet like some of the early
founders of Sun, NetApp, et. al., and I'm still a working stiff, I
lead a comfortable life, and it seems like a good tradeoff to me. :-)
In the long run, it might be interesting to see how the Illumos (Open
Solaris) derivatives fare compared to Free/Net/Open/Dragon BSD's.
There seem to be some interesting cooperation from the set of
companies that use Illumos, which is encouraged by the CDDL's weak
provisions. So if Illumos and its derivatives are able to overtake
*BSD's despite the *BSD's having an earlier start, that might be an
interesting confirmation of dwheeler's point above.
- Ted
7:51 p.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
On Mon, Jan 20, 2020 at 2:10 PM Theodore Y. Ts'o <tytso(a)mit.edu> wrote:
.... which is dated July 1992, and describes a
"launch" of 386BSD
Ouch.... this is why people get confused. I'm probably not helping by not
being precise enough.
386/Net/Free/OpenBSD as >>projects<< are different from the original 386
work that Bill and Lynn did to get it running on the first 386 and get that
code back into the CRSG project. The 386 was released in 1985 by Intel.
A number of firms started to use it. Wikipedia says Compaw was first (The
Compaq Deskpro 386/25 was '89 according to google and we used a Wyse 32:16
at Stellar before that). Bill Jolitz did is work originally on a Multibus
based 386 from Intel (using his NS system to bootstrap) and moved it to the
PC (I thought a Deskpro) shortly after it started to ship.
Most of us call that work '386BSD' but it was 4.X BSD that booted on a 386
based PC and not really what you are discussing. These bits are/were the
'hidden' ftp - which is not to be confused with the first BSD 'distro'
for
the 386 which would be much more public,
BTW: The original BSD on a 386, install was very rough. Bill had created
the boot floppies on his NS system. You had a use a DOS program to create
them via an image copy and the boot was really funky. As I said, the
original AT driver was wrong, and kept getting hosed until I fixed it when
I was consulting for NCR (on their 386 system in 1989). IIRC Jolitz had
created it by looking at that Minx AT driver and made some guess. I had
(think I still have) the WD1003 documents, so I knew what the registers
really looked like and it was not handling some error conditions IIRC.
And the naming is really the root of the whole argument BTW.... BSD 4.x
for a 386 based system *vs.* a real distribution. Bill Jolitz tried to
make a better release for a BSD on a 386 (*a.k.a.* 386BSD) The install for
a PC/386 improved. IIRC, Intel had paid CMU to do some work as part of
Andrew and the Mach stuff (Bob Barron was the author of this I believe).,
They wrote a version fdisk, and a bunch of things to allow dual booting
and some other tools that ran on DOS. I don't remember how, but Bill
Jolitz got that code and very early on the BSD 386 port used it - probably
as part of the CMU/Mach to CSRG push/intermixing.
And later on yet to Jordan's credit, and it was after Linux, NetBSD, *etc*
was all there, that FreeBSD, completely redid the install scripts and made
the system that pretty much is the model for all current PC based systems
now. In fact, around that time I had started work with Linux and one of
the things liked about FreeBSD 1.x was the install compared to the
original Linus package (although I had an early Slackware pretty earlier
in my Linux time and that improved things). I think your comment and about
the healthy competition was true, each team was trying to do better.
11:04 p.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
At Mon, 20 Jan 2020 14:09:00 -0500, "Theodore Y. Ts'o"
<tytso(a)mit.edu> wrote:
Subject: Re: [TUHS] Early Linux and BSD (was: On the origins of Linux - "an academic
question")
On Sun, Jan 19, 2020 at 07:32:57PM -0800, Greg A. Woods wrote:
.... which is dated July 1992, and describes a "launch" of 386BSD
Release 0.0 in March 17, 1992. This is contemporaneous with Linux
0.95a (which by coincidence was also released on March 17th, 1992.)
Yes, though as I recall all of the articles mentioned that the OS could
be downloaded, but I pointed at that final article as it was the first
one in which I found explicit mention of the FTP server name(s).
Out of
curiosity, did the articles contain download information for a
bootable copy of 386BSD?
Yes, they did:
https://www.drdobbs.com/porting-unix-to-the-386-the-final-step/184408800 The first "real" distribution, the Soft
Landing System, was released
in May 1992. (The Manchester Computer Centre distribution in November
1991 was a floppy-based distro containing command-line and development
utilities, but not X Windows, so some people don't feel it counts as a
full-featured distribution.)
The actual 386bsd Release 0.0 (the one done directly by Bill and Lynne
Jolitz) announcement is dated "March 7, 1992" according to the first
post about it on comp.unix.bsd (and according to that announcement there
was a meeting at Apple in Cupertino (SVNet) on the 11'th where copies of
the floppies were made available for copying (comp.unix.bsd:
<2763(a)tardis.Tymnet.COM>)M>).
Note that according to an article from Unigram ("Issue 396", dated
August 3-7, 1992, (p)re-posted by Tom Limoncelli in comp.unix.bsd) this
"386bsd 0.0" was actually a re-write of earlier work to create a
"386"
based release of BSD. Apparently UCB lawyers asked Jolitz to destroy
all the initial work done for the release, and he complied and rewrote
what became 0.0 from scratch again, starting with the plain NET2
release. (comp.unix.bsd: <1992Aug1.020513.14170(a)plts.uucp>)
I would argue that in one way of looking at things NetBSD (and by
extension FreeBSD) really started with the 0.0 patch kit, and that's
also dated March 15, 1992 by Chris Demetriou. I agree though that the
creation of the first commits in the CVS repository represent a more
direct reflection of the intent to create a unique thing called NetBSD.
(On March 13, 1992 there was a post by Mike Stump on comp.unix.bsd
asking for someone to coordinate patches for 386bsd; and Pace Willisson
posted the first patch in response on March 14, 1992; and Chris replied
on the same day saying he would put such patches up on
agate.berkeley.edu; and the "README.PATCHES" file appeared there on
March 15, 1992.)
--
Greg A. Woods <gwoods(a)acm.org>
Kelowna, BC +1 250 762-7675 RoboHack <woods(a)robohack.ca>
Planix, Inc. <woods(a)planix.com> Avoncote Farms <woods(a)avoncote.ca>
21 Jan
21 Jan
12:13 a.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
On Mon, Jan 20, 2020, 4:06 PM Greg A. Woods <woods(a)robohack.ca> wrote:
At Mon, 20 Jan 2020 14:09:00 -0500, "Theodore Y.
Ts'o" <tytso(a)mit.edu>
wrote:
Subject: Re: [TUHS] Early Linux and BSD (was: On the origins of Linux -
"an academic question")
Lots of people were building CVS repos based on the patchkits... Chris
wasn't trying to start a project, but more was trying to find a way of
organizing everything that people were working on. At least that's what I
recall from the rumors I'd heard on campus after Chris visited Boulder...
Warner
(On March 13, 1992 there was a post by Mike Stump on comp.unix.bsd
On Sun, Jan 19, 2020 at 07:32:57PM -0800, Greg A. Woods wrote:
> > Out of curiosity, did the articles contain download information for a
> > bootable copy of 386BSD?
>
> Yes, they did:
>
>
https://www.drdobbs.com/porting-unix-to-the-386-the-final-step/184408800
.... which is dated July 1992, and describes a "launch" of 386BSD
Release 0.0 in March 17, 1992. This is contemporaneous with Linux
0.95a (which by coincidence was also released on March 17th, 1992.)
Yes, though as I recall all of the articles mentioned that the OS could
be downloaded, but I pointed at that final article as it was the first
one in which I found explicit mention of the FTP server name(s).
The first "real" distribution, the Soft
Landing System, was released
in May 1992. (The Manchester Computer Centre distribution in November
1991 was a floppy-based distro containing command-line and development
utilities, but not X Windows, so some people don't feel it counts as a
full-featured distribution.)
The actual 386bsd Release 0.0 (the one done directly by Bill and Lynne
Jolitz) announcement is dated "March 7, 1992" according to the first
post about it on comp.unix.bsd (and according to that announcement there
was a meeting at Apple in Cupertino (SVNet) on the 11'th where copies of
the floppies were made available for copying (comp.unix.bsd:
<2763(a)tardis.Tymnet.COM>)M>).
Note that according to an article from Unigram ("Issue 396", dated
August 3-7, 1992, (p)re-posted by Tom Limoncelli in comp.unix.bsd) this
"386bsd 0.0" was actually a re-write of earlier work to create a
"386"
based release of BSD. Apparently UCB lawyers asked Jolitz to destroy
all the initial work done for the release, and he complied and rewrote
what became 0.0 from scratch again, starting with the plain NET2
release. (comp.unix.bsd: <1992Aug1.020513.14170(a)plts.uucp>)
I would argue that in one way of looking at things NetBSD (and by
extension FreeBSD) really started with the 0.0 patch kit, and that's
also dated March 15, 1992 by Chris Demetriou. I agree though that the
creation of the first commits in the CVS repository represent a more
direct reflection of the intent to create a unique thing called NetBSD.
asking for someone to coordinate patches for 386bsd;
and Pace Willisson
posted the first patch in response on March 14, 1992; and Chris replied
on the same day saying he would put such patches up on
agate.berkeley.edu; and the "README.PATCHES" file appeared there on
March 15, 1992.)
--
Greg A. Woods <gwoods(a)acm.org>
Kelowna, BC +1 250 762-7675 RoboHack <woods(a)robohack.ca>
Planix, Inc. <woods(a)planix.com> Avoncote Farms <woods(a)avoncote.ca>
11:45 p.m.
New subject: Early Linux and BSD (was: On the origins of Linux - "an academic question")
At Mon, 20 Jan 2020 17:13:48 -0700, Warner Losh <imp(a)bsdimp.com> wrote:
Subject: Re: [TUHS] Early Linux and BSD (was: On the origins of Linux - "an academic
question")
Lots of people were building CVS repos based on the patchkits... Chris
wasn't trying to start a project, but more was trying to find a way of
organizing everything that people were working on. At least that's what I
recall from the rumors I'd heard on campus after Chris visited Boulder...
Indeed, that is no doubt a far more accurate way of stating things.
I was only watching from afar, effectively.
Of course it didn't take too long before he announced 0.8 and made it
all official.
--
Greg A. Woods <gwoods(a)acm.org>
Kelowna, BC +1 250 762-7675 RoboHack <woods(a)robohack.ca>
Planix, Inc. <woods(a)planix.com> Avoncote Farms <woods(a)avoncote.ca>
18 Jan
18 Jan
3:30 p.m.
On Fri, Jan 17, 2020 at 10:50:51PM -0500, Theodore Y. Ts'o wrote:
+1 to everything Ted said, that's how I remember it as well. I knew
about both Linux and 386BSD and while 386BSD felt very familiar to
a SunOS guy, there was something special about Linux. For a while I
played with both, 386BSD was sort of better in that it had networking,
but like Ted, my network was a modem and TCP over a modem wasn't pleasant.
So Linux won out and eventually networked just fine.
At the time when Linus announced his creation (not yet
named) on
comp.os.minix in August 1991, it was already self-hosting. And that
happened pretty quickly; he first started working on the project in
June or July.
Around the end of 1991, I had added Job Control (implemented from
POSIX.1 as a the specification), so we could put jobs in the
background. In 1992 X Windows was ported to Linux. Networking
support followed shortly thereafter.
--
---
Larry McVoy lm at mcvoy.com http://www.mcvoy.com/lm
So all in all.. As I remember it, there was never
really a decision to 'make
this great new OS!'.. It kinda happened with right place, right time, right
people, etc.
In the super-early days (late 1991, early 1992), those of us who
worked on it just wanted a "something Unix-like" that we could run at
home (my first computer was a 40 MHz 386 with 16 MB of memory). This
was before the AT&T/BSD Lawsuit (which was in 1992) and while Jolitz
may have been demonstrating 386BSD in private, I was certainly never
aware of it --- and Linus was publishing new versions every few days
on an ftp site. We'd send patches, and in less than a week, there'd
be a new release dropped that we could download.
So the argument, "Linus would have never started on Linux if itT
weren't for the AT&T Lawsuit" I don't think fits with the timeline.
Development was very fast paced, and so it was *fun*. And at least
for me, the lacking of networking during the early days didn't bother
me much, since I didn't have networking at home. (I didn't have
grounded outlets, either, in my 3 people for $1050/month apartment.
Each leg was 50-60V to ground, and the wiring was cloth wrapped, and
was either steel or aluminum; I never did figured out which....)
Using zmodem over a 2400 bps modem was way more efficient than PPP, so
even once we had networking, I didn't always bring up pppd. And the
most common way I would download source was using set of 1.44 MB
floppies and a station wagon (literally; I was driving a Corolla wagon).
During those early days, the fact that Linux was more "primitive" than
BSD may have been an advantage, since it sources was small, and
release engineering is simple when you only support one architecture.
The other things I noticed was that because we didn't have the weight
of the Unix/BSD legacy, we were more free to experiment. Bruce Evans
was working on the serial driver for FreeBSD, and I was working on the
serial driver for Linux, and we had a friendly competition to see who
could get better throughput using the very primitive 8250 and later
16550 UART. The figure of merit we were using was the CPU overhead of
a C-Kermit file transfer over two RS-232 ports connected via a
loopback cable. We'd compare notes to see how we could make things
better, me for Linux, and Bruce for FreeBSD, and it was *fun*.
Eventually, it got to the point where I was making changes to the tty
layer to further optimize things, and at that point Bruce reported
that he couldn't do some of the optimizations, since it would have
required changing the TTY layer that had been handed down from the
Gods of Olympus^H^H^H^H^H^H^H^H BSD and so it was nixed by his
colleagues in FreeBSD land.
In contrast, in Linux, people felt free to rip out and replace code if
it would make things better. Depending on how you count things, the
networking layer in Linux was ripped out and replaced three or four
times in the space of as many years. Sure, the first version was
pretty crappy, and was barely good enough for simple telnet
connections. But things got better fast, because people were felt
free to experiment.
My personal belief is that it was this development velocity and
freedom to experiment starting with a super simple base is what caused
Linux to become very popular amongst the those who just wanted to play
with kernel development. Compare and contrast Linus's willingness to
accept patches from others and his turnaround time to get those
patches into new releases with Bill Jolitz's 386BSD effort --- and I
don't think you need the AT&T lawsuit to explain why Linux took off in
1991-1992. FreeBSD and NetBSD was started in 1993 because of the
failure of Jolitz to accept patches in a timely fashion.
- Ted 
17 Jan
17 Jan
11:11 p.m.
On 1/17/20, Arrigo Triulzi <arrigo(a)alchemistowl.org> wrote:
The answers I got varied from “the world needed a free Unix and BSD was
embroiled in the AT&T lawsuit at the time” to “Plan 9 also had a restrictive
license” (to the latter my response was that “so did Unix and that’s why
Linus built Linux!”) but I don’t feel any of the answers addressed my
underlying question as to what was wrong in the exposure to other operating
systems which made Unix the choice?
Linus has always struck me as purely a pragmatist and not idealistic
at all, so I'm not surprised that he wrote a conventional Unix rather
than something more architecturally progressive.
On 1/17/20, Brantley Coile <brantley(a)coraid.com> wrote:
Plan 9 solves the problem of "How do I make a bunch of machines look like a
single system?" If you wanted to mess around with a system in the early
1990's you didn't have a bunch of people and a bunch of systems you needed
to make appear as one. You just had a single box.
So, my Plan 9 remains small. In fact, I've been removing things from it,
like local disks, that is contrary to the original vision. (Or set of
visions. I remember getting a lot of different answers form everyone
involved in 1127 about what it was that they were doing.)
Wasn't the point of single-system-image clustering originally to allow
building relatively inexpensive systems with more CPUs than could
reasonably be fit into a single machine? Now that all current CPUs
except for some low-end embedded ones are multi-core and fully
programmable GPUs are ubiquitous, I don't think Plan 9/Amoeba-style
SSI is really all that relevant for anything other than HPC. However,
I do think distributed network-transparent sharing of devices and
services along the lines of QNX or Domain/OS is more relevant than
ever.
18 Jan
18 Jan
12:23 a.m.
There's a book called "Just For Fun: The Story of an Accidental
Revolutionary" co-authored by Linus Torvalds and David Diamond that
sets out how he came to write the Linux kernel.
Wesley Parish
On 1/18/20, Arrigo Triulzi <arrigo(a)alchemistowl.org> wrote:
[I originally asked the following on Twitter which was
probably not the
smartest idea]
I was recently wondering about the origins of Linux, i.e. Linux Torvalds
doing his MSc and deciding to write Linux (the kernel) for the i386 because
Minix did not support the i386 properly. While this is perfectly
understandable I was trying to understand why, as he was in academia, he did
not decide to write a “free X” for a different X. The example I picked was
Plan 9, simply because I always liked it but X could be any number of other
operating systems which he would have been exposed to in academia. This all
started in my mind because I was thinking about my friends who were CompSci
university students with me at the time and they were into all sorts of
esoteric stuff like Miranda-based operating systems, building a complete
interface builder for X11 on SunOS including sparkly mouse pointers, etc. (I
guess you could define it as “the usual frivolous MSc projects”) and
comparing their choices with Linus’.
The answers I got varied from “the world needed a free Unix and BSD was
embroiled in the AT&T lawsuit at the time” to “Plan 9 also had a restrictive
license” (to the latter my response was that “so did Unix and that’s why
Linus built Linux!”) but I don’t feel any of the answers addressed my
underlying question as to what was wrong in the exposure to other operating
systems which made Unix the choice?
Personally I feel that if we had a distributed OS now instead of Linux we’d
be better off with the current architecture of the world so I am sad that
"Linux is not Plan 9" which is what prompted the question.
Obviously I am most grateful for being able to boot the Mathematics
department’s MS-DOS i486 machines with Linux 0.12 floppy disks and not
having to code Fortran 77 in Notepad followed by eventually taking over the
department with X-Terminals based on Linux connected to the departmental
servers (Sun, DEC Alpha, IBM RS/6000s). Without Linux they had been running
eXeed (sp?) on Windows 3.11! In this respect Linux definitely filled in a
huge gap.
Arrigo
1555
days inactive
1560
days old
48 comments
21 participants
participants (21)
-
Andrew Warkentin -
Arno Griffioen -
Arrigo Triulzi -
Arthur Krewat -
Bakul Shah -
Brantley Coile -
Clem Cole -
David Barto -
George Michaelson -
Greg 'groggy' Lehey -
Greg A. Woods -
Jon Forrest -
Jon Steinhart -
Larry McVoy -
Lars Brinkhoff -
reed＠reedmedia.net -
Rob Pike -
Theodore Y. Ts'o -
Warner Losh -
Warren Toomey -
Wesley Parish