Amiga OS 3.1 Source code leaked yesterday

[kamelito]

The leak was way older than yesterday, it was spread during the Amiga technologies era, but now it is widespread...Totally illegal I guess.
Kamelito

Originally Posted by dlfrsilver

it's maybe the complete source. I got access to it, and the files are full of "Amiga Confidential" notifications in the C source files.

Only a coder can say. Contact Toni Wilen, and let him check. He will say if it's complete or not.

EDIT : there is no word about Cloanto, so the archive has been leaked by one of the original amiga engineers, there's no other possibility.

Original Amiga engineers... pure speculation.
Kamelito

[Korodny]

Like kamelito already said: this stuff hasn't been leaked now, it's been leaked in the late nineties. Rumors have it that Tyschtschenko handed out the sources to a whole bunch of people in Germany (in addition to Phase5, which already had them) to make sure they don't get lost during the next bankruptcy.

These sources were all over the warez boards in 1997 or 1998.

It's the original 3.1 sources from Commodore, they require quite an exotic setup to compile (an obscure SunOS/Solaris based Cross-Compiler amongst other things, IIRC), Olaf Barthel had to spend quite a few hours to make them compile/assemble with modern tools before they could be used as the base for OS 3.5+.

[Locutus]

Ofcourse most didn't have access to those warez boards in the 90s, so its still news for a lot of people of these being easily accessible now.

[gulliver]

I would like to receive a PM

Originally Posted by Korodny

Like kamelito already said: this stuff hasn't been leaked now, it's been leaked in the late nineties. Rumors have it that Tyschtschenko handed out the sources to a whole bunch of people in Germany (in addition to Phase5, which already had them) to make sure they don't get lost during the next bankruptcy.

These sources were all over the warez boards in 1997 or 1998.

It's the original 3.1 sources from Commodore, they require quite an exotic setup to compile (an obscure Solaris based Cross-Compiler amongst other things, IIRC), Olaf Barthel had to spend quite a few hours to make them compile/assembly with modern tools before they could be used as the base for OS 3.5+.

Wow, had no clue. I guess a lot of people didn't know. I read upon the story and there were some compilers used (Bell Labs C compiler” (i.e. PCC) , and this 3.1 source code was cleaned up a lot before they used it for 3.5+. "It's a mess" was the word about this code. So I guess it's not that fun, as a cleaned up 3.1+ had been. Still, interesting enough.

[dlfrsilver]

Originally Posted by kamelito

Original Amiga engineers... pure speculation.
Kamelito

maybe

[Olaf Barthel]

Originally Posted by morpa

Wow, had no clue. I guess a lot of people didn't know. I hope Olaf documented his compile job somewhere...

Well... if you put it like that.

It's much more than a "compile job". If you looked at the archive contents (which I do not condone) you would find that the Amiga operating system consists of some 280 (give or take) individual components (a component would be a library, a device driver, a shell command or application program such as "HDToolBox").

Each component has to be built individually, and in many cases with a particular compiler (Green Hills, Lattice 'C', SAS/C, Manx Aztec 'C'), or a particular assembler, or it would either fail to build correctly, or at all (e.g. only the Green Hills compiler can build a correct intuition.library).

Originally Posted by kamelito

Original Amiga engineers... pure speculation.
Kamelito

Having had a look at the archive I would say that this theory has its merits. None of the files in the archive are newer than May 5th, 1994. Commodore went bancrupt in late April 1994.

To me this suggests that somebody with access to both the AmigaOS 3.1 source code and other material from the same server, as well to material from the Commodore Amiga Technical Support (CATS) department, must have used the last opportunity to grab it.

[kamelito]

@Olsen,
It could have been saved for not being lost, that doesn't mean that it has been leaked on purpose by one of the original team.
Is that material different for the one you saved from DAT backup?
Kamelito

Originally Posted by Olaf Barthel

Well... if you put it like that.

It's much more than a "compile job". If you looked at the archive contents (which I do not condone) you would find that the Amiga operating system consists of some 280 (give or take) individual components (a component would be a library, a device driver, a shell command or application program such as "HDToolBox").

Each component has to be built individually, and in many cases with a particular compiler (Green Hills, Lattice 'C', SAS/C, Manx Aztec 'C'), or a particular assembler, or it would either fail to build correctly, or at all (e.g. only the Green Hills compiler can build a correct intuition.library).

Interesting, It must have been a hell of a job/challenge getting the code up to respectable standards. I just read some old interview where it was stated as a mess, and I can see why if this was the case. What was the hardest parts besides a mountain of individual obscure compilers and components, if your'e allowed to say? Thanks for hearing a part of the story.

[Olaf Barthel]

Originally Posted by Korodny

Like kamelito already said: this stuff hasn't been leaked now, it's been leaked in the late nineties. Rumors have it that Tyschtschenko handed out the sources to a whole bunch of people in Germany (in addition to Phase5, which already had them) to make sure they don't get lost during the next bankruptcy.

I heard these rumours, but as far as I can tell from the contents of today's leaked archive, this leaked archive is much older than the material which Amiga Technologies GmbH received in 1995.

In fact, what Amiga Technologies GmbH received was the complete set of files, as available then. Today's leaked archive only contains a small portion thereof. I daresay that this leaked archive comes from a completely different source.

Originally Posted by kamelito

@Olsen,
It could have been saved for not being lost, that doesn't mean that it has been leaked on purpose by one of the original team.
Is that material different for the one you saved from DAT backup?
Kamelito

Yes. Today's leaked archive contains a small subset of the material provided to Amiga Technologies GmbH back in 1995.

[kamelito]

Originally Posted by Olaf Barthel

I heard these rumours, but as far as I can tell from the contents of today's leaked archive, this leaked archive is much older than the material which Amiga Technologies GmbH received in 1995.

In fact, what Amiga Technologies GmbH received was the complete set of files, as available then. Today's leaked archive only contains a small portion thereof. I daresay that this leaked archive comes from a completely different source.

Meaning that work was still done on the OS after CBM bankruptcy?
Kamelito

[Olaf Barthel]

Originally Posted by morpa

Interesting, It must have been a hell of a job/challenge getting the code up to respectable standards. I just read some old interview where it was stated as a mess, and I can see why if this was the case. What was the hardest parts besides a mountain of individual obscure compilers and components, if your'e allowed to say? Thanks for hearing a part of the story.

The hardest part was in testing and retesting the build over and over and over again.

I was hardly the first person to attempt and succeed at getting AmigaOS 3.1 to build again (that honour goes to Andy Finkel). I was sufficiently crazy to try and get it to build completely native on AmigaOS, using the smallest set of compilers and assemblers (specifically, without having to resort to using the Green Hills compiler).

That goal necessitated changes in the code (hence the retesting), and reconstructing a common build environment which allowed you to open a shell, enter "make" and wait for a couple of hours until everything had been rebuilt from scratch. This, as far as I know, had not been possible since 1986.

[kamelito]

@Olsen
Was it automation testing or manual testing?
How can be sure that everything was working as before?
Kamelito

[Olaf Barthel]

Originally Posted by kamelito

Meaning that work was still done on the OS after CBM bankruptcy?
Kamelito

No, I did not mean to imply this.

The simplest explanation for today's leaked archive to me is that a Commodore engineer made a copy of this material for personal reasons. This kind of thing happened and still happens all the time. In some case it's a good thing, where material which would otherwise be lost could be preserved for future use.

Given how much trouble Commodore was in before the end I think it is very well possible that whoever made the copies thought of it as rescuing what he could from a smoking and in other parts blazing library.

Originally Posted by kamelito

@Olsen
Was it automation testing or manual testing?

Strictly manual testing only (on an A3000, A1200 and an A600HD bought second-hand for the occasion). This was a private project of mine after all, in the hope that it might become useful, which it eventually did starting in 1997/1998.

I did not learn about automated testing until a lot later, and even if I had know about it back then it would have been a lost cause. AmigaOS is not designed and implemented in a manner which would facilitate testing.

[kamelito]

@Olsen,
How long did it take from start to finish to?
Kamelito

[Olaf Barthel]

Originally Posted by kamelito

How can be sure that everything was working as before?

Well, you can't. That would have required an extensive testing effort on a variety of hardware configurations which I did not have access to.

What I did, however, was to make the operating system components build using almost exclusively using the same tools which the last working AmigaOS 3.1 build in 1993/1994 would use. Where it made sense to switch build tools (say, from Manx Aztec 'C' to Lattice 'C') because the impact of the change was very small, I made the change. Changes in, for example, the printer driver build procedure were less critical than in changing the compiler needed to build graphics.library, for example.

In so many words, I tried to limit the impact of the changes and when I made changes, I made small changes only.

The big problem was with intuition.library which had to be ported from K&R 'C' (ca. 1985) to ANSI 'C'. That is a different story altogether.

Originally Posted by kamelito

@Olsen,
How long did it take from start to finish to?
Kamelito

Let's see... according to the CVS logs I must have worked on it for 3-4 years, on and off.

[kamelito]

@Olsen,
That is a lot of time even if not fulltime, thanks for the work done.
Cya
Kamelito

[Leffmann]

Someone has already pointed out that the source code for SetPatch and other small things are missing from the archive, so it's definitely not complete.

@Olaf Barthel, somewhere there is probably a lawyer-type running around with his pants on his head, screaming like a crazy monkey about this, but realistically, what do you think will be the outcome of this?

[dlfrsilver]

Originally Posted by Olaf Barthel

No, I did not mean to imply this.

The simplest explanation for today's leaked archive to me is that a Commodore engineer made a copy of this material for personal reasons. This kind of thing happened and still happens all the time. In some case it's a good thing, where material which would otherwise be lost could be preserved for future use.

Given how much trouble Commodore was in before the end I think it is very well possible that whoever made the copies thought of it as rescuing what he could from a smoking and in other parts blazing library.

Seems i was right after all

[kamelito]

Call your wife, play the lottery
Kamelito