OSNews

Ehm. It's me, or this guy is completely unaware of the existence of ReactOS? It would have been the most obvious OS featured in this article.

I was thinking the exact same thing, I was like "WTF mate?"

A fair point but I had to be selective in terms of which operating systems I looked at or the article would have been even longer.

I am quite familiar with Amiga and OS/2 and that eased the research stage. In addition, I had some points that I wanted to make about those OSes.

I might do a follow up at some point featuring BEOS, WIN NT, etc inspired projects.

"What about Linux recreating Unix ?"

Linux was more directly a recreation of Minix.

Linux was more directly a recreation of Minix.

Which was really just a unix-like implementation for x86 right?

Really, BSD would probably be one of the best candidates for "recreation of unix"

"Which was really just a unix-like implementation for x86 right?"

Sure, that's why I said "more directly." Minix, Linux, BSD, OS X, etc. are all descendants of Unix and can probably be considered "recreations."

But, IIRC, Linus' original purpose for Linux was specifically to recreate Minix because the Minix license wouldn't allow him to extend it.

//But, IIRC, Linus' original purpose for Linux was specifically to recreate Minix because the Minix license wouldn't allow him to extend it.//

That may or may not have been the purpose, but Linux is fundamentally different from Minix.

They work and are structured in very different ways. Linux is no a "recreation" of Minix ... more like a complete replacement.

I think that this is one of the points that I was trying to make - the distinction between OSes which are externally similar and OSes that are internally similar. This holds true for Minix also (see further down thread). Minix and Linux are externally quite similar but *internally* they are implemented differently.

Mike

"Really, BSD would probably be one of the best candidates for "recreation of unix""

Correct me if I am wrong, please, but BSD was not a recreation of Unix. It IS Unix.

BSD was developed at Berkeley based on the Bell Labs project. Bell's Unix included the source code. Berkeley built on this and over the years because BSD.

Bell Labs (or AT&T) and Berkeley went to court over who owns BSD. They settled out of court....

Nice little diagram:
http://upload.wikimedia.org/wikipedia/commons/5/50/Unix_history-sim...

Neat diagram! Although I got dizzy and nearly vomited from the confusion

According to Twenty Years of Berkeley Unix, you are wrong. ( http://www.oreilly.com/catalog/opensources/book/kirkmck.html )

"During one of our weekly group meetings at the CSRG, Keith Bostic brought up the subject of the popularity of the freely-redistributable networking release and inquired about the possibility of doing an expanded release that included more of the BSD code. Mike Karels and I pointed out to Bostic that releasing large parts of the system was a huge task, but we agreed that if he could sort out how to deal with reimplementing the hundreds of utilities and the massive C library then we would tackle the kernel. Privately, Karels and I felt that would be the end of the discussion.

Undeterred, Bostic pioneered the technique of doing a mass net-based development effort. He solicited folks to rewrite the Unix utilities from scratch based solely on their published descriptions. Their only compensation would be to have their name listed among the Berkeley contributors next to the name of the utility that they rewrote. The contributions started slowly and were mostly for the trivial utilities. But as the list of completed utilities grew and Bostic continued to hold forth for contributions at public events such as Usenix, the rate of contributions continued to grow. Soon the list crossed one hundred utilities and within 18 months nearly all the important utilities and libraries had been rewritten.

Proudly, Bostic marched into Mike Karels' and my office, list in hand, wanting to know how we were doing on the kernel. Resigned to our task, Karels, Bostic, and I spent the next several months going over the entire distribution, file by file, removing code that had originated in the 32/V release. When the dust settled, we discovered that there were only six remaining kernel files that were still contaminated and which could not be trivially rewritten. While we considered rewriting those six files so that we could release a complete system, we decided instead to release just what we had. We did, however, seek permission for our expanded release from folks higher up in the University administration. After much internal debate and verification of our method for determining proprietary code, we were given the go-ahead to do the release."

Regarding UNIX amd OSRs, "The UNIX system family tree: Research and BSD" is very interesting. You can find it on any FreeBSD installation or live system CD (FreeBSD, DragonflyBSD, FreeSBIE, PC-BSD, DesktopBSD etc.) in /usr/share/misc/bsd-family-tree or http://www.freebsd.org/cgi/cvsweb.cgi/~checkout~/src/share/misc/bsd... (text file with ascii diagram).

Really, BSD would probably be one of the best candidates for "recreation of unix"

Not really. Linux is more Unix-like than BSD.

//Linux was more directly a recreation of Minix.//

Err, no.

Linux was written on a Minix computer, but Linux is nothing like Minix.

Minix is a microkernel, Linux is a monolithic kernel, for a start. Fundamentally different from the get go.

http://en.wikipedia.org/wiki/Minix
"MINIX is an open source, Unix-like operating system based on a microkernel architecture. Andrew S. Tanenbaum wrote the operating system to be used for educational purposes; MINIX also inspired the creation of Linux. Its name derives from the words minimal and Unix."

http://en.wikipedia.org/wiki/Linux_%28kernel%29
"Architecture

Linux is a monolithic kernel. Device drivers and kernel extensions run in kernel space (ring 0), with full access to the hardware, although some exceptions run in user space. The GNU/Linux graphics subsystem (the X Window System) is not part of the kernel, is optional, and runs in user space, in contrast with Microsoft Windows.

Kernel mode preemption means device drivers can be preempted under certain conditions. This latter feature was added to handle hardware interrupts correctly, and to improve support for symmetric multiprocessing (SMP). Preemption also improves latency, increasing responsiveness and making Linux more suitable for real-time applications.

The fact that Linux is not a microkernel was the topic of the Tanenbaum-Torvalds debate[12] which was started in 1992 by Andrew S. Tanenbaum with Linus Torvalds regarding Linux and kernel architecture in general on the Usenet discussion group comp.os.minix.[13] Tanenbaum argues that microkernels are superior to monolithic kernels and that, for this reason, Linux is obsolete. This subject was revisited in 2006.[14] [15]

Unlike traditional monolithic kernels, device drivers are easily configured as modules, and loaded or unloaded while running the system."

Chalk and cheese.

OS/2 almost never crashed as it features Protected Memory since the MS OS/2 1.X era.

Ran it for many years... Crashed? I can only remember once or twice. Lockups? All the f*%@ckin time until they fixed the SIQ, which didnt happen till Warp was a few years old. Is there a difference from the users perspective? Nope, sux either way. I did like the UI though - that and it ran 16-bit windows and DOS apps _really_ well.

Never had many lockups. Seen a few traps but never that many.

Well to be fair the SIQ itself didn't cause lockups, buggy software did. My experience with lockups was more due to the software I was trying to run. In fact if you were a run-of-the-mill OS2 user back in the day, the odds are that the only OS2 native apps you ran were WPS and maybe a few utilities. Most people ran Windows and DOS apps to be honest. And if you didn't run native OS2 apps you would never run into the SIQ issue.

Anyway, if I remember correctly the 2 programs that caused the most trouble were Footprint Works (later IBM Works after IBM bought it for the release of Warp) and early versions of Lotus Smartsuite for OS2.

There was something about Mozilla (circa 0.6) that seemed to make it trap for me. Also, there seemed to be a black art about finding exactly the right fixpack level for a stable machine.

I had some lockups with OS/2 2.1 - but that was probably due to my TSENG ET4000 graphic card (you know - the first serie generation - the one with the red screen bug).

It ran wonderfully smooth with only 4 MB of RAM. It could even "play" video (well... "fast" moving still-images

I don't understand why he mentions drive letters. Windows uses drive letters too (and drive labels as well).

In regard to WPS I have to say that the template function and the general object oriented approach is something which should really be implemented in other OS'es. ASAP! (Gnome actually have some OS/2-like support for templates - it is just hidden away).

Edited 2006-11-22 20:39

An even nicer feature of OS/2 was the "shadow copy" feature. Symlinks and shortcuts on steroids. You could make a shadow copy of an app, then move the original app, and the shadow would still point to it. Edit the shadow and you could change features of the original.

Something I've missed in every other OS I've tried since.

Yup, "a better DOS than DOS". I think it was John C. Dvorak who said that.
Warp was actually quite usable, Merlin (v4) kicked ass for it's time.
Anyone interested in it can go look for the eComstation LiveCD, that gives you an idea.

Just out of curiosity, why would you consider BSD an OSR of Unix? Considering that the code has evolved from the original Unix rather than being written from scratch (which is the case of Linux and many real OSRs), it doesn't make much sense. Also, in terms of ideas, System V Unix has taken as much from BSD as it has given them.

Edited 2006-11-22 22:40

"Just out of curiosity, why would you consider BSD an OSR of Unix?"

Because it is. It was rewritten from scratch for licensing reasons. Since Networking Release 2, all of six original files were left over, and even those were replaced after some years.

As it stands, there is no Unix code in any of the BSDs.

I agree totally.
In my opinion, it is not fair to compare any current operating system with the OS/2 Version 4 from 1996, because since then a lot of fixpaks (17 in total!) have been published by IBM with tons of updates, bugfixes and new features:
http://www.warpupdates.mynetcologne.de/english/base_fixpak40.html
E.g. Fixpak 15 is freely available since 2001 and updates your Warp 4 installation up to version 4.50 which is very different from its predecessor. So it would be fair and necessary to compare other (current) operating systems at least with this more modern 2001 version of OS/2 Warp 4.
And, moreover, please do not forget Ecomstation, the modern successor of OS/2 (OS/2 Warp Version 4.52). In my opinion, a comparison with this latest "Warp 4" would be the best.

I have to agree. The author seems like he needed something to write about and didn't do much research. I guess you could say that eComStation is to OS/2 as ZETA is to BeOS, only properly executed. eCom is a pretty big jump up from what Warp 4 is/was. My dad still runs it exclusively. Has no issues with new hardware. So what if he can't play Duke Nukem Forever (hehe), he is actually productive on his machine. He has Thunderbird, Firefox, OpenOffice, lots of imaging apps, etc. On my installation of eCom, I run VirtualPC and can run WinXP, BeOS, and tons of linux distros, thereby giving me the power of the apps from ALL those OSes.

For a reference, I run eCom on an IBM NetVista X41 P4 1.8 Ghz, 1 GB RAM, 160GB HD, and a 16x DVD+/-RW, USB2, Hauppauge PVR-150. There's other goodies in there, but base system wise it's pretty recent. I have PVR ability, which is something linux won't even do on that machine, much less BeOS and ZETA which followed. WinXP likes to hiccup during recording.

So I'd say it's a pretty damn useful OS in it's current form, especially considering it can do things that "current" OSes can't.

Not to sound like a rant (sorry if it did) but I really feel more research would have gone to make this article better than it was.

I think you're wrong.

Linux can do it with the right drivers(it can), as could beos with the right drivers(no clue).

I used a ati aiw radeon on linux and beos on a pIII 550 to watch tv, and record (rarely because of driver issues)

Hardware capture cards don't need a power processor to run.

For a reference, I run eCom on an IBM NetVista X41 P4 1.8 Ghz, 1 GB RAM, 160GB HD, and a 16x DVD+/-RW, USB2, Hauppauge PVR-150. There's other goodies in there, but base system wise it's pretty recent. I have PVR ability, which is something linux won't even do on that machine, much less BeOS and ZETA which followed. WinXP likes to hiccup during recording.

Linux has PVR ability and in fact your machine is probably fully supported under Linux. Intel hardware has better support under Linux than most other hardware and the hauppauge 150 is also supported. You could run MythTV on that with no problem.

I think you forgot this...
http://ati.amd.com/products/certified/genesi.html (PPC, Linux, ATI/AMD, etc.)
...and this:
http://www.genesippc.com/press.php?date=20050624 (and that was some time ago)

"The bad is that the writer hasn´t been in touch with OS/2 in a long time. IBM did update Warp in 1999 and included USB support for most devices, JFS file system, kernel upgrades to support more memory, a proto 32 bit device driver model, updates to the TCPIP stack with new functions to accelerate Web work (web servers)... etc "

It is true that I have been out of the os/2 scene for while. However, I stand by what I said:

"Perhaps IBM could have kept OS/2 relevant but they didn't make any serious efforts to develop it beyond about 1996."

I could perhaps have been a bit clearer in my definition of 'serious efforts to develop' but I consider 'development' to be different to 'maintenance'. Perhaps, they were serious about the platform for a bit longer than 1996 too. My understanding was that IBM were forced to maintain OS/2 in order to comply with their corporate contracts. By the time I left the platform, IBM were recommending that customers transition over to windows/Linux.

New printer drivers, the deal with scitech for graphics support and USB support kept the platform usable on new hardware. However, I consider that to be different to a 'serious effort to develop' the OS.

I didn't even knew AmigaOS was "inspired" on Unix*...

*it ain't

Edited 2006-11-23 00:07

More precisely, it was based on TriPOS (written in BCPL)

One more way is a safe/managed language. So programs can run safe without any memory protection at all

Imagine a language that need to check if every byte is safe to read or write :-p

A simple code like this:

int main()
{
a=startmem;
for (;*a!=0;a++;) *a=20;
}

Become quite completed and slow.

int safe_read(a)
{

For (i=0;i<tabsize;i++)
{
If ((Memtab[ i ].startmem=>a) and (a<Memtab[ i ].endmem))
{
Return *a;
}
}
Memory_exeption();
}

void safe_write(a,data)
{

For (i=0;i<tabsize;i++)
{
If ((Memtab.startmem=>a) and (a<Memtab[i].endmem))
{
*a = data;
Return;
}
}
Memory_exeption();
}

int main()
{
a=startmem;
for (;safe_read(a)!=0;a++;) safe_write(a,20);
}

Now just imagine the extra needed assembler that need to be executed for just calling a safe_write() and safe_read(), safe languages like basic are NOT where popular whit programs that needs to be fast like games for example.

[i]Edited 2006-11-24 16:30

Imagine a language that need to check if every byte is safe to read or write :-p

It is not as bad as you think :
- compiler can rearrange code to move checks out of cycle. So the code you provided will be as fast as "not safe" version.
- programs can be preverified to eliminate redundant checks.

It is better to have 5% slower, but stable program.

>And without memory fragmentation BTW, which is a one of the nastiest Amiga problems

That’s way AmigaOS4 has new memory allocation system integrated too, that supports virtual address space.

I think that we may have to agree to disagree on this point. I still maintain that IBM threw in the towel on the OS for political rather than technical or economic reasons. The OS still had something going for it at a point when IBM was advising its customers to pursue Windows, Linux and other Unix instead of OS/2. I still stand by my point that even the other extras that you mention do not constitute 'serious development of the platform by IBM'.

Also, let's not forget the original focus of the article: I was arguing against the re-implementation of OS/2. I am not arguing against people, such as yourself, getting pleasure/utility out of sticking with OS/2. Do you feel that a project to recreate OS/2 would be a good use of resources?

I don't have a problem agreeing to disagree. :-) If the world all agreed with each other, it's be a boring place to live...

OS/2, as a general desktop OS, has several qualities that I have yet to find to the same extent in Linux, the BSDs, or Windows, or BeOS, or ... , and that has been enough to keep me using OS/2 to this point.

I like the mix of a 4OS2/4DOS command set (DOSlike but far less limited), REXX, a filesystem which retains case without being case-sensitive, low resource usage, a high level of responsiveness to changing process loads, and a very high level of out-of-the-box legacy DOS and Win16 app support along with a more modern 32-bit native API.

For me, the WPS is a nice thing, but it isn't the reason I like and use OS/2. It's not even near the top of the list. I'm a fullscreen console guy who likes scripting and aliases, not a GUI clicker. :-)

I'd certainly like to see it stick around in some form. As I see it, there are two ways to try and keep it alive:

(1) Try to prop up the existing kernel(s) with a newer supporting cast of device drivers, installation code, useful utilities, and other things, or

(2) Try to recreate the best aspects of OS/2 (hopefully including some level of native OS/2 API support) in a new OS built from scratch.

#1 is doable now, and probably provides the best bang for the buck, but the fact that the kernels are static now (or relatively so) and binary-only makes long-term survival a question.

#2 would ensure long-term viability, but OS/2 is not a simple operating system (it's single-user, but it has a lot of APIs and uses some sophisticated x86 features), and the fact that it has more than one native API and an extensible OO desktop makes the task of re-creation a nontrivial one.

My own vote is for #1 as long as possible, with #2 going on in the wings. :-)

I do think recreating it would be a very good use of resources. Why? Because it would provides one of the very few non-UNIX-like alternative platforms which actually has enough software to be a viable general desktop solution, and I think that would appeal a lot more to Windows folks than either Linux or BSD would.

OS/2 is not like Windows, not really, but it's closer to the Windows way of thinking than Linux is without making many of the security/UI/resource sacrifices that Windows is saddled with. I see it as a happy medium which would otherwise not exist.

Edited 2006-11-28 22:09