Commodore UNIX, some questions

[olsen]

The last time I checked on this, WinUAE does not support an 030 with the MMU which AMIX needs.  I too think it would be cool to boot AMIX on WinUAE but Toni has kind of nixed the idea.

This seems to have changed with the most recent WinUAE release (2.6.0). I can't tell if it actually did (got a full plate right now), but once I find the time, I'm going to make disk images of my Amiga Unix machine and give it a spin.

I used AMIX in college on 3000 UX and 3000 TUX machines.  The 3000 TUX even had the 3070 tape drive mounted inside which was pretty cool.

Gee, they actually shipped A3000T units with Amiga Unix? I always wondered about that.

AMIX (Unix SVR4) was pretty amazing for the day.  Dual booting was pretty radical at the time and most people who saw it were suitably impressed.  However, AMIX is a pretty old Unix compared to today.  Xwin is B&W (Without the 24bit 2410 card) and it does not support DHCP among a whole list of other things.  Of course, emulating the 2410 board would be pretty easy I would think and that would give Xwin color under WinUAE.

Um, emulating the A2410 (TIGA graphics hardware) may not be such a hot idea after all. If I remember correctly, the X server which supported the card was X11R4 and it suffered from memory leak bugs. Source code was unavailable, which made fixing the bugs difficult.

One might have better luck with a Picasso II emulation and installing the matching X11R5 server, but I'm speculating: this is still really hard to do.

[olsen]

I have a (sort-of) insite on this subject. UNIX v1-7 (called "Ancient UNIX") are now open source, I'm running the SIMH PDP-11 emulator on a FreeBSD machine. It is running a clean install of UNIX v6 that was installed from a rip of a PDP-11 tape of the UNIX 6 distro. I don't think Amiga UNIX, like that of UNIX past v7 is free.

Ownership of the product has changed several times, too, which complicated things enormously.

Back when I was doing consulting work for Amiga Technologies GmbH we made an effort to obtain the Amiga Unix source code (which was not available among the assets acquired by ESCOM). It could have been useful

Searching through the available debris, I figured that it might be possible to contact the Amiga Unix authors who had ported the AT&T code to the Amiga, and written the Amiga-specific drivers. It turned out that one of these authors was still around, did respond to e-mail and was willing to help. The big catch was that for legal reasons he had to make sure that Amiga Technologies GmbH actually was legally entitled to receive the source code in question. This turned out to be practically impossible to prove, due to the big ball of mess that was the Commodore asset acquisition by ESCOM.

[olsen]

I am no lawyer, but wouldn't the copyright return to the programmer once the company he wrote the software for went bankrupt? Especially since no one else bought the rights to it.

Amiga Unix shipped with the source code to the Amiga-specific bits, and as far as I know these were never ever lost. You could rebuild the kernel, for example, changing the Amiga-specific code, but the portions which were not Amiga-specific were provided as object files in binary form only.

The portion we were after was the AT&T System V Release 4 source code itself, including Amiga-platform specific patches. In a nutshell, we would have liked to be able to rebuild the entire Amiga Unix kernel, userland files and everything, and turn Amiga Unix back into a commercial product.

Problem was, this wasn't a 1980'ies game, but some serious, very expensive code base which was valued north of US$ 500,000 at the time.

Well, it probably was a pipe dream to be able to upgrade the kernel code (support the 68040, support more peripherals such as DAT drives, for example). But we tried anyway, and we failed

But, thankfully, NetBSD finally came around at the time when we tried to revive Amiga Unix. It would have been very difficult to revive Amiga Unix, get it back into shape, since it was completely closed source. Given the state of the Amiga market at the time, I doubt that enough customers would have been willing to pay for upgrades (especially since Commodore did their best to burn any bridges it may have still had to the Amiga Unix user community).

[olsen]

I see, thanks for the details, it's a very interesting story.

Have you ever thought about writing an historical book about your experiences? I'd buy it, in fact I'd even pay for a pre-order!

Ha! Here comes my Forrest Gump moment

I did not keep a journal in these days, and what I recall cannot be the complete picture. Let's hope that http://www.amiga.org will be archived for future generations to wonder what that thing "Amiga" is everybody's talking about in 2317.

[olsen]

Speaking of source code, does anybody have the source code to AMIX?

It's possible. Whoever I managed to contact back in 1995, when I tried to get the source code back for Amiga Technologies GmbH, did appear to me to be the kind of professional programmer who would hang onto such things, in case somebody came calling (just like I did with the AmigaOS source code). Still, as the years go by, the knowledge to make the code work again does tend to fade.

Whoever that enity is, can't we ask for a copy of the source code?

You can always ask, but you can't always have what you want. It's probably easier to stir the pond and wait for the original AT&T SVR4 source code to rise to the surface than to try and do this for the Amiga Unix source code. I bet that many more people had access to the AT&T SVR4 source code than were ever involved with the exotic Amiga Unix source code.

In 2013, I'm not sure how you could make money off of it.

That's easy for you to say. Think "lawsuit". Specifically, think "IP lawsuit": http://en.wikipedia.org/wiki/SCO_Group

The last rumor I heard was that the source was accidentally destroyed by Gateway when they reformatted some media (HD/Tapes).

Yes, they probably threw the wrong switch at the same time and accidentally disintegrated larger parts of South Dakota, too (which has gone largely unnoticed even today).

As far as I know, Commodore sold off its hardware assets (workstations, PCs, Amigas, manufacturing machines) when it went into liquidation. Before the computers were sold, however, the contents of the hard disk drives, etc. were stored on magnetic tapes (e.g. QIC). Some of those tapes were quickly made available to ESCOM, and they did contain the almost entire Amiga business material (operating system source code, manuscripts, software, etc.). But some of the backup tape contents never surfaced again. The content listings I saw suggest that these tapes were mostly backups of CAD workstations. So, it's possible that some data was lost, but the Amiga stuff is still around and in really decent shape (probably better shape than it was in 1994).

As far as the Amiga Unix source code is concerned, I suspect that it may have never been stored on the Commodore computers which were still around when the backup tapes for ESCOM were made. AT&T licensed the Unix source code to manufacturers and it would not surprise me if the licensing agreement stipulated security measures intended to limit the exposure the source code got within the company which ported it to their hardware. Remember, back in 1989 there was just one commercial grade Unix solution, and it was worth a lot of money to company which owned it. Part of that value came out of keeping it closed source.

For that matter, who has the source code to Workbench and Kickstart?

I know somebody who knows...

 Is it the same people?

It's possible, but somewhat unlikely.

[olsen]

IIRC the Amiga port was not done (completely) by Commodore, but by UniSoft. The same company which ported SVR4 to the Atari TT/Falcon and Macintosh (called A/UX.)

Interesting; I didn't know that there were so many 68k ports of SVR4 around. By 1992 Unix usage had started to tilt towards RISC platforms (SPARC and MIPS), and Motorola's 68040 did not exactly draw a crowd (although Apollo Computer/NeXT used the 68040 in a number of models).

This company specialised in ports to 68k architectures.
The reason why the source code of A/UX, AMIX and Atari is not available is that (large) parts of it still belong to AT&T. Part of the reason why A/UX, AMIX and Atari UNIX didn't get much support is that the royalties were huge.

Yes, that would explain why Commodore got out of the market, at breakneck pace. If anything, the management were cheapskates. Still, given the cost of the A3000 platform, it might have made a viable low-end Unix workstation had Commodore had more trust in building a business around it. Other entry-level workstations available in 1989/1990 were certainly much cooler, and there were whole "ecosystems" already in place which they fit into (e.g. Sun), but you had to pay a lot more money for this gear.

In the end, Apple at least decided to acquire NeXT and develop it in-house.

But then NeXTSTEP was a pretty exotic Unix to start with. It had 4.2BSD user code/kernel APIs, which sat on top of the MACH kernel. Back in 1995 it was already tough to compile and build current GNU/Unix code on a NeXT machine. Bootstrapping gcc became next to impossible, and even something comparatively simple such as building ssh would break your heart...

If I remember correctly, Apple ended up rebuilding the kernel from the inside out, using parts of FreeBSD. The userland software part of NeXTSTEP (the entire ObjC framework which lives on Mac OS X) was worth the price, but not so much the kernel underpinnings.

A pity, since A/UX is pretty cool.

I can't say I've ever used it. But it it's SVR4 compatible, it might actually offer better support for building "current" software on it than NEXTSTEP wound up.

[olsen]

As a University Professor who taught IP, Copyright, and Patents for seven years, I know all about this.  :-)

Sounds tough, given how ripe with side-effects the field has become over last century, and in particular the past decade.

However, I think you misunderstood what I was saying...  I'm not sure the owners of AMIX (source code and all) could make any money off of it from a product stand point.  It runs on such limited hardware, would have such a limited market, and is so outdated that it is nearly useless.

That may be true, but such an analysis seems to be done very rarely. Even if it is done, it appears that nobody wants to explore the consequences that may arise from releasing the property.

For example, somebody must have believed that extending copyright protection for works produced by a corporation to 120 years would be a good idea. But what class of works still possesses value after 120 years? Speculatively, there might be some value left, so letting it go (even if it currently appears to have no discernable value at all) might result in a loss of revenue, so it's easier to decide against letting it go than to choose the alternative.

I guess we'll find out in the year 2048 how much Disney still makes off "Steamboat Willie", when it might finally go into the public domain  If it is allowed to go, it could mean that there is no practical value left. Seriously, how much value is there to 16mm black and white cartoons with sound effects after 120 years?

And you're talking about operating system source code which is barely 25 years old. Somebody might not want to risk releasing such a valuable property right now, as it could arguably still generate revenue.

On a side note, the SCO example would not apply to this particular situation.  I could explain why but it is long and complicated.

I would expect so, and would want to add "sad" to the list of applicable adjectives, too.

Actually, I think AMIX would be hard to defend for IP violations.  
-P

Could be, but I bet there's going to be somebody who'd champion to defend the cause anyway, and win, too  It is easier to say "no" to this for the property owners than to say "yes".

But, given that you have the necessary background, why not give it a try and ask Novell? That might be an interesting exercise (and produce material for coursework)