What is dead may never die: a new version of OS/2 just arrived An outfit called Arca Noae has released a new version of IBM's venerable OS/2 operating system, named ArcaOS 5.0. The Register understands that Arca Noae has a licence from IBM to do a distribution of OS/2, the OS that Big Blue pitched against Windows 95 back in the day. OS/2's fourth release was widely regarded as technically …
"Has it got a TCP/IP stack and internet browser as well? yes and yes"
OS/2 had both of those in 1994. More import is that it supports USB, which wasn't on OS/2 or Win95 in 1995.
I remember MIcrosoft OS/2 Server with LAN Manager (sold in 1989 after IMB and MS parted ways).
Is OS/3 the reason that NT starts at Version 3?
MS inflicted Win9x, not much more than Win3.x with the VFW & Win32s bundled optimised for gaming on businesses that would have been far better using NT3.5 or OS/2 Warp. A burden for them and business till XP (Win 5.1, Win2K was the unfinished Win 5.0). Idiots, though commercially sucessful, it was the source of most of the stupid design decisions and badly written SW on NT after 1995.
I think Microsoft started its version numbering for Windows NT with 3.1 because that was the same version number of the Windows desktop OS at the time. Both desktop and server OS largely shared the same GUI interface.
I remember installing Windows NT 3.1 from floppies for one of my college professors. Due to lack of software support, it wasn't exactly dazzling. It did offer a TCP/IP stack, a trick that took Novell a few (fateful) years to replicate.
<blockquote>
I think Microsoft started its version numbering for Windows NT with 3.1 because that was the same version number of the Windows desktop OS at the time. Both desktop and server OS largely shared the same GUI interface.
I remember installing Windows NT 3.1 from floppies for one of my college professors. Due to lack of software support, it wasn't exactly dazzling. It did offer a TCP/IP stack, a trick that took Novell a few (fateful) years to replicate
</blockquote>
The software that was released as Windows NT was started [pre divorce] as OS/2 NT - the 'portable' version of OS/2, written in C instead of assembler. It also had a kernel + API design [Win32, Win16, OS/2 1.x, POSIX subset]
From what I recall, NetWare 3.x was introduced a couple of years before WinNT - and had several pluggable network protocols - both TCP/IP and IPX were available, but IPX was easier to configure.
From what I recall, NetWare 3.x was introduced a couple of years before WinNT - and had several pluggable network protocols - both TCP/IP and IPX were available, but IPX was easier to configure.
Technically that's accurate, though the TCP/IP implementation under NetWare 3.x, 4.0, and 4.1 was quite limited (enabling compatibility with local FTP, LPR, and HTTP, but no routable communications). The implementation under 4.11 was laughable (encapsulating TCP/IP in IPX... lolwut). NetWare did get a native implementation in 5.0, but by that time many customers had abandoned NetWare.
"I think Microsoft started its version numbering for Windows NT with 3.1 because that was the same version number of the Windows desktop OS at the time. Both desktop and server OS largely shared the same GUI interface"
Seem to recall there was also somethjing about using this numbe so MS was able to claim that NT was just another version of Windows 3 which avoided some contractual undertakings they'd made with IBM over they future roadmap being OS/2.
NT statrted at 3.1 as it was contemporary to Win3.1 at the time. Effectively the business version of it, much like 95/98 had NT4, and Me had Win2K.
OS/2 came about because the original was an improvement over the original DOS (and initially could only run text based applications, albeit in 32bit) and they touted it as the new DOS (hence OS/2). That and the IBM PS/2 line.
"MS inflicted Win9x, not much more than Win3.x with the VFW & Win32s bundled optimised for gaming on businesses that would have been far better using NT3.5 or OS/2 Warp. A burden for them and business till XP (Win 5.1, Win2K was the unfinished Win 5.0). Idiots, though commercially sucessful, it was the source of most of the stupid design decisions and badly written SW on NT after 1995."
A lot of people who are now rabid Linux fanboys and general MS haters (or was at some point, like me in my younger days) switched to Linux from Win95/98 and still think "Windows" is like that
.
Win95* was a pretty darn impressive hack with a lot of sheer sorcery needed to build an environment that's simultaneously a new 32-bit system with all the bells and whistles, still very backwards compatible with 16-bit Windows and DOS stuff (even to the point of supporting DOS-only hardware drivers!) and had good performance on the systems of the day.
But this came at great costs in terms of stability, functionality and elegance. Which people without an understanding of the underlying reasons behind it - and more importantly, having never tried NT in any meaningful way - quickly assumed was because of "M$" incompetence when writing their "Windoze" (after all, Linux was clearly better, and also made you feel superior to common users after mastering it or even managing to install it and getting X running in high res).
So if they had just started pushing NT sooner, maybe we would have less annoying Linux fanboys...
Still, I'm not sure I can blame them for not doing it at that point in time. I really think the 9x stop-gap was needed to get people to start developing the shiny new Win32 applications, and NT wasn't exactly nice with 8 or even 4MB RAM which were common then.
They should probably have skipped Win Me though and simply launched XP at that point in time...
* Actually the biggest change from a technical viewpoint was arguably when Windows got 32BFA, which arrived in the 3.11 minor release, but...
"But this came at great costs in terms of stability, functionality and elegance. Which people without an understanding of the underlying reasons behind it - and more importantly, having never tried NT in any meaningful way - quickly assumed was because of "M$" incompetence when writing their "Windoze""
Just because NT was a proper OS, it doesn't follow that Win 95 was OK. There is a reason Ctrl-Alt-Delete is a famous key combination. MS can sometimes do things properly, then the right small group of people work on a project -but as a rule, business decisions trump technical decisions. Just witness the things that never get fixed, because management has decided that it won't be cost effective. Despite billion dollar profits.
Maybe you should try reading what I wrote again?
It was "OK" as a stop-gap measure to stick 32-bit preemptive multitasking on top of DOS and Win3.1x on the systems of the day.
Not "OK" compared to "proper" 32-bit systems that didn't have to give a damn about such things as backwards compatibility.
I can recommend reading the book Unauthorized Windows 95 if you think the result was caused by incompetence or business decisions, as opposed to hard technical limitations regarding what can be achieved under those circumstances.
" if you think the result was caused by incompetence or business decisions, as opposed to hard technical limitations"
You could argue that just sticking Windows 95 on top of DOS was a business decision. A competent business decisions (made money), leading to an incompetent technical decision (gave us the Ctrl-Alt-Delete meme).
In many organisations the techies would have pushed back with "we aren't doing that because ugly/stupid/support costs, we are doing Windows 2000+ instead, aka XP". At MS techies were more compliant it seems.
"A lot of people who are now rabid Linux fanboys and general MS haters (or was at some point, like me in my younger days) switched to Linux from Win95/98 and still think "Windows" is like that"
I doubt that very much. Do you think they have not noticed XP, Vista, 7 and 10?
XP came out 16 years ago! And if they didn't like 95/98 they most likely knew about Windows NT and 2000 as well.
Linux stands on it's own merits. Not on believing Windows is still Windows 95.
> "Has it got a TCP/IP stack and internet browser as well? yes and yes"
> OS/2 had both of those in 1994.
Yes ... OS/2 2.0 (or possibly 2.1) was promoted as being the "on-ramp to the internet" and came with the ability to sign up for the ibm.net ISP. It also had this amazing (at the time) feature that when they issued a service pack you could download it over the internet and it automatically installed itself (so no need to find 10+ floppies to copy the images onto and then shuttle through the floopty drive!).
>businesses that would have been far better using NT3.5 or OS/2 Warp.
I tried.And it's not like I was tied to Windows at the time. Didn't like Win3. OS/2 user long before Win95. Painful Win95 installations, and it wasn't enormously stable.
But Win 95 application availability was better, and development support was better, and that's where I ended up, because overall, it was far better than using NT3.5 or OS/2 Warp.
If IBM had used the 68008 rather than the 8088 the PC may have started a teeny bit slower but we would have avoided 10 years of some of the finest brains in the world trying to work out which 64k block their brain had exploded in.
Shit even MS might have written a decent OS on that!
The Intel segmentation model is actually quite sophisticated. Too bad that MS doesn't really use it -- sure, it segregates code into 'user' and 'system' rings but prefers to put everything in a nice big memory puddle where applications can corrupt each other, data can be run as code and so on. MS went this route because a lot of their earlier GUI code was based on what Apple was doing and Apple was using the 68K processors, early versions of which lacked memory segmentation.
This is just about the weirdest misconception I have ever heard...
The Apple GUI code was Pascal and 68k ASM.
The Windows (NT) GUI code is C and C++ (plus a tiny bit of ASM for the various archs supported, none of which is even remotely similar to 68k).
The non-NT Windows GUI code is C and C++ as well, plus quite a bit of ASM (x86).
The reason for the flat memory model and lack of any finer separation than user/kernel is that NT was always intended to be portable - they had to go with the common denominator across all relevant archs with MMUs. In fact, NT wasn't even originally developed for x86 and the x86 port was done pretty late in the development of the first version.
That being said, segmented protected mode and multiple rings certainly have a point (apart from scaring youngsters about the horrors we had to endure).
Atleast earlier in its life (1.x, 286 16-bit protected mode) OS/2 used this with quite good results.
SO it's actually based on the OS/2 code-base or is it a dark-room clone? It's not entirely clear to me from the article.
A lot of my late 90's and early 2000's were spent working on OS/2. My first intro was 2.1 where Iearnt just how annoying it was loading the 21 or so disks to get the OS up and running.
I have many fond memories - including how we could run Windows apps on it natively not only more stably, but faster than they ran on a Windows machine. Quite often apps that would crash consistently on Windows would just run smooth as silk under OS/2. Pretty impressive for an emulated environment.
Deary God No !!!!!
I suffered OS/2 the first time round and it needs a few more stakes thru the heart for everyones sake !!!
Leave it be and let it 'rest in piece' ....... lots of them :)
If OS/2 comes back then I want Netware to come back as well as at least it worked ..... very well in my experience, so there.
[Before the Anti-Netware people arrive :)]
It was not I could not use the OS, it was that it was unstable for totally random reasons.
Software could run well for a time then stop for no reason with no meaningful error messages.
Two machines could be built with the same media and software installed that would run on one but not the other.
(This was on identical kit sourced from our warehouse and tested for faults etc before use as standard. Including IBM, HP, Compaq and lesser known brands of the time.)
This was for a large Distributor/Services company that shipped thousands of systems to customers that they supported. They had no lack of very good skilled people to do the work who were trained properly by the likes of IBM, Netware, Microsoft, HP, Compaq etc. The company used the hardware/software they sold for real on their internal systems, not just slinging out 'kit' to customers.
The stability was the issue as I remember, I had issues logged with IBM that where never resolved !!!
(This was many years go so don't ask for details I thankfully forget the specific details.)
Finally, as per your 1st line .... I am speaking for my self and my experience and the experience of some our our customers at the time.
Perhaps, after all this time I need to 'Let it go !!!' but some traumas stay with you. :)
Edit timed out on previous reply ..... :(
(Part 2)
Forgot to mention.
A chromebook or similar would not suffice for the needs of the Customers I supported, I built and installed & supported mid-large systems based on AIX, HP-UX, Solaris, Netware and even a little VMS thrown in for luck. they were intergrated into networks that supported PC based workstations, of the time (which I also configured, built & installed) and strangely enough I had no real problems that could not be solved. I could build/configure all these systems from the hardware configuration upwards and 100% working as specified including OS and supported software.
Just maybe OS/2 was not as good as the 'Marketing' .... (not an unknown event in IT, I would say !!! :) )
I will now let this go.
It was a difference in design philosophy. A lot of problems that were blamed on OS/2 were actually hardware that wasn't performing to spec. Stodgy OS/2 said "I'm not going to stand for substandard hardware." Windows said "if I didn't see it, it didn't happen". The odds ended up in Windows favor and the underlying issues went unnoticed. But they had their share of blue screens too.
OS/2 was designed to run perfectly on IBM machines (esp. the PS/2 ones) - running it on anything else was pretty much Russian Roulette because it looked up specific things only found in IBM BIOSs so was 80% reliable at best on non-IBM kit... if it ever installed (it doesn't like Toshiba laptops that much).
"running it on anything [but IBM machines] was pretty much Russian Roulette because it looked up specific things only found in IBM BIOSs" ...
I am quite skeptical of this. I ran it for a time on self assembled '486 and dual Pentium Pro systems. Neither one had an IBM BIOS, and neither one had noticeable problems with OS/2. Unlike Windows of the time, it was rock solid. It did give problems with VMs on qemu/kvm when tried a few years ago.
I installed and/or ran OS/2 on many different machines, so far as I can remember not a single one of them an IBM-branded unit. If you could install it, it just worked. Weeks, months and years of uptime - and that in an era when nothing else except 'nix could manage more than a few days at best. Windows didn't start to approach OS/2 stability until 2000 came along. Consumer-grade Windows never did.
"OS/2 was designed to run perfectly on IBM machines (esp. the PS/2 ones)"
Not really. I once spend a whole weekend trying to get OS/2 Merlin to work on an IBM Aptiva and it simply didn't work. No OS/2 drivers provided, no native support, it was a total failure.
Funny thing too: Compaq actually provided OS/2 drivers for some of their hardware, but not IBM.
Niltze [Hello]!
IBM Aptiva computers required these upgraded drivers -- overwriting the originals in either diskette 1 and/or diskette 2:
ibm1s506.add
ibmidecd.flt
os2cdrom.dmd
os2dasd.dmd
cdfs.ifs
Evidently, a professional installer had to add corresponding entries in CONFIG.SYS (found in diskette 1):
[...]
basedev=ibm1s506.add
basedev=ibmidecd.flt
device=\os2cdrom.dmd
basedev=os2dasd.dmd
ifs=cdfs.ifs /q
[...]
Rem AND set at the end:
SET COPYFROMFLOPPY=1
Cheers!
@Mage
Domino 6.5.5 runs a treat on 2k8 R1 x86 (i was bored and wondered) :P and as hard as it may believe, Notes 6.5.4 runs sweet as a nut on W7 x64 (its never been so stable, no more ZapNotes required)...I have the (mis)fortune of still supporting it and being certed in Dom.....sure you have to grant all users R/W to Data folder and notes.ini but its never been so stable....
Tell router update config.... :P reboot schmeeboot
Odd - I never had any issues with our OS/2 Notes servers. They just ran and ran, and were fairly nippy too.
Now, the Notes server on a Netware box running as an NLM - let's just say that replacing that was a top priority, and much beer was consumed when it happened. Netware 3.x was wonderful for file & print services, but godawful for running application servers on.
"Odd - I never had any issues with our OS/2 Notes servers. They just ran and ran, and were fairly nippy too."
The only problems I ever had running Notes on OS/2 was when the Notes Admins did something dumb. Like turning off the machine because shutdown took too long (because Notes was doing it's shutdown).
A few corrupted os2.ini and os2sys.ini files later the Notes Admins weren't allowed in the data center any more.
I think it was a missed opportunity that they didn't put a compatibility layer in there for Linux or BSD. Maybe something like built-in VM support and an integration layer with X for the desktop. One of the big problems with OS/2, and the reason for the solid support of Win16 apps, is the fact that there were so few native applications. I know I used it primarily for multitasking DOS applications that had performance problems with Desqview or Win95.
There was EMX, gcc, gdb and XFree86 all on OS/2 providing source level compatibility with *nix apps.
I ran a number of X apps on OS/2 and even used it to write *nix/X apps before moving the source across the network to HPUX and building and testing on that target platform.
OS/2 was a great platform running multiple OS's from DOS, Windows, *nix, Java besides OS/2 apps.
IBM screwed up when they got OS/2 running full Win32 apps before Microsoft shipped Windows 95. Once Microsoft found out, they changed the SDK resource compiler such that it put a tiny amount of resources at the top end of their virtual address space. That was up at 4GB and OS/2 could only access 2GB of virtual address space so there was no way for OS/2 to run these Win32 apps any more. OS/2 stuck running just Win16 and Win32C apps. And with no compatibility there wasn't a migration path.
... hardware using things like USB that had come along since OS/2's last version, 4.52, emerged in 2001.
A bit of revisionist history here. USB1 came out in 1996 (with some silicon available at least a year before that). USB 1.1 dates back to 1998, while USB 2.0 was released/ratified in 2000/2001. By 1999, it was already getting hard to find new PC hardware which did not include a USB port in some form.
By now, USB is a fair bit older than many of its users; the thing is practically ancient!
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
