I have a strong feeling that OS/2 2.0 (besides requiring a 386) would have been a lot more ambitious than Windows 3.0 would been

[-]

VivienM7@reddit

And part of the problem with ambition was the hardware requirements... remember, this is an era where 4MB of RAM costs $250CAD.

Windows 3.1 and later 95 could run 'well enough' with less RAM than OS/2 (and let's not talk about NT, which needed a minimum of 12 megs at a time when 4 was common).

(That being said, there was a period in late 1994/early 1995 when people were excited about OS/2 Warp. Didn't last too long...)

[-]

richardsequeira@reddit

One thing I always remembered about OS/2 was that it was for the PCs that people wanted. Windows ran on the PCs that we all had.

[-]

yuhong@reddit (OP)

Note that OS/2 1.x was however limited to 16MB of RAM.

[-]

-jp-@reddit

That was a pretty opulent amount of memory at the time anyway. 16,384k ought to be enough for anyone. :)

[-]

Malice_Qahwah@reddit

Heh. I have two complete OS2 warp install disk box sets. Still in shrinkwrap iirc. Been saving them for when I get around to building a 90s era PC

[-]

richardsequeira@reddit

Well to keep in mind for OS/2 Warp 3, you are looking at the minimum of 27 disks. The complete set can be a total of 37 disks with utilities.

[-]

Malice_Qahwah@reddit

No idea I've not opened them! They're still sealed.

[-]

Fluffy-Queequeg@reddit

I had OS/2 2.1 and then OS/2 Warp running as a triple boot setup on my 486DX/50 in 1993/1994. Other two partitions on my then massive 1.3Gb HDD were a Dos/Win3.1 setup and a Linux Slackware setup.

I used to run Windows 3.1 from within OS/2 a lot of time. I did have native OS/2 apps but they weren’t well supported and it was mostly done as part of Computer Science degree

[-]

bmiller218@reddit

Galactic Civilizations (Gal Civ) used multi threaded code for the AI.

[-]

Fluffy-Queequeg@reddit

If I recall, it was basically the Lotus office suite and a C/C++ compiler and a terminal emulator

There really wasn’t much available, and it was more of an academic thing for me. Even Windows was only used for Word/Excel.

[-]

I'm kind of jealous of that even. I caught the OS/2 bug in 1998 when I had a DX2 hand-me-down to tinker with; I never had any productive use for it, even though I liked exploring this "OS from another dimension"

[-]

Fluffy-Queequeg@reddit

I was lucky. I was living in Japan immediately before going to university, so I built up my PC in Tokyo. When I arrived in Australia they weren’t selling 486DX/50 systems. Most students were buying 386 systems or maybe a 486SX/25 and then I show up with this high end 486 tower and a ridiculous 8Mb of RAM and Opti Local Bus video.

[-]

glwillia@reddit

excellent DOS/Windows support was both OS/2’s strong point and its fatal flaw. since windows apps ran so well, nobody bothered to make native OS/2 apps

[-]

nderflow@reddit

Your question is phrased as a hypothetical but OS/2 version 2.1 actually shipped (presumably v 2.0 did also but I never used that).

[-]

yuhong@reddit (OP)

I am also thinking in terms of MS getting involved in the project as well.

[-]

TableDuck@reddit

I think this is an interesting what if, and there are several points where OS/2 2.0 could have been “the” PC OS.

If we look at the OS/2 1.x books it was considered fact (when they wrote it) that Windows 2.0 was just a stepping stone for what would be an OS/2 centric PC landscape. In every one of those books it was always when, not if, we are all running OS/2.

There was the 386 issue that Gates & Co pushed pretty hard for IBM to abandon the 286, and refund their customers for orders placed for 286 machines. Similar to developers groaning over having to have Xbox Series S support in Series Xbox game they ship. IBM pushed back, and it led to some very impressive programming by Gordon Letwin and his team for OS/2 1.x to switch from protected mode to real mode.

Then we have the Dave Cutler problem. Dave and his DEC team join Microsoft in 1988, and reject all of Letwin’s/IBM’s work to date for what was going to be OS/2 3.0 - later NT. This could have been a real collaboration, where you could have had a real fusion between the two (three, if you include Letwin’s team). Cutler and his team would have pushed for the OS to be C, not assembly based, leading to more portability for IBM’s RISC based systems. It also would have opened the door for a UNIX like operating system on PC for consumers, harking back to the DOS 2.0/Xenix dream that Bill Gates and Paul Allen wanted for a moment.

Had I been aware of OS/2 2.0 at the time (it didn’t get a lot of press in the Ziff-Davis world) - I would have snatched it up in a second. DOS, Windows, and 32bit OS/2 program support in a real protected mode OS? Sign me up.

[-]

dpdxguy@reddit

I am also thinking in terms of MS getting involved in the project

MS was involved in OS/2. It was originally a joint project between Microsoft and IBM. MS pulled out when Gates realized that having their own OS would give MS a whole lot of leverage over the x86 computer marketing.

If I remember right, by the time of OS/2 2.0 or 2.1, MS was still involved, but they were sabotaging IBM's efforts.

[-]

yuhong@reddit (OP)

Of course, I am talking about OS/2 2.0 here.

[-]

Web-Lackey@reddit

OS/2 2.0 would not have been more ambitious had Microsoft been fully engaged. it actually would have been less ambitious. Microsoft was opposed to the Workplace Shell: they thought that using the Windows 3.0 / Program Manager interface was fine, and that 100% of the effort should be put into building a 32-bit operating system.

You know, exactly what they did with Windows NT? :-)

And if you look through the history of OS/2 2.0 betas/DDK/whatever releases, both from IBM and from Microsoft, you will see that they were 100% correct. The Workplace She’ll added something between a year and two years of development time. It caused the memory needed for a usable system to just about double, and it added very little to the actual productivity of the system.

Don't get me wrong: I loved me some Workplace Shell: it was incredibly powerful. But in a world where 2 MB was a decent amount of RAM and 4 MB was top-end hardware, requiring 8 MB in order to run well…. all that did was further push possible adoption of OS/2 out another two years (to 1994) when systems caught up to it. And by then, Chicago (Windows 95 code name) was front and center, taking even more mindshare…

Source: running OS/2 since 2.0 LA (about four months before GA!), and having bought a literally top of the line 486 DX-33 (*before* there was such a thing as a DX/2) with 4MB of RAM (in early 1992) which cost $2400, only to have to immediately spend another 4MB so that the system wouldn’t have to swap *just* to get the desktop to appear….. (RIP a PS/2 Model 70 that could run OS/2 1.3 just *fine* with 4MB…)

[-]

yuhong@reddit (OP)

I am of course not talking about the Workplace Shell here. I am talking about other features that was being planned.

[-]

Web-Lackey@reddit

Well, then, I can simply say this: you would be wrong. OS/2 2.0 would look the same. Even before the split, Microsoft wasn’t working on OS/2 2.0. IBM was. Microsoft was working on OS/2 3.0 (called OS/2 NT at the time). After the split, that was simply renamed Windows NT.

So we know what the answer is. OS/2 2.0 would look like OS/2 2.0, and OS/2 3.0 *might* have looked like Windows NT 3.1 but with WPS. I say ”might” because IBM had a terrible habit of forcing poor choices on Microsoft. So it’s likely that had IBM and MS stuck together, we still might have gotten something much closer to Warp 3.0 than NT 3.1.

But I’ll re-ask my question: what do you think we might have gotten? Why would you think we would have gotten something more “ambitious” than OS/2 2.0 or NT 3.1? IBM was *not* exactly known for “ambitious” in the *entire* history of the IBM PC, and I don’t think that MS staying partnered with IBM would have made them *more* ambitious than what they actually were in this timeline.

(I might call the PS/2 and MicroChannel “ambitious”. They certainly were more capable. But the extra cost, coupled with the obvious attempt to make the IBM PC architecture more proprietary was not ambition for the benefit of its users…)

[-]

Gadgetman_1@reddit

OS/2 2.1 was my first experience with a 'Server OS'. All the PC servers at the office ran OS/2 2.1 with Lan Manager 2.2 The interface was a bit cludgy, but it worked. No frills, just the essentials.

Going to winNT 3.51 was a huge downer. Literally. Uptime was NOT included in that package...
Also, who the F! includes CPU-heavy screensavers on a Server distro?

OS/2 Warp 3.0 just... clicked with me. Man, when I ran it on my Pentium 66, with 64MB RAM, and a SCSI Quantum FireBall 1.3GB HDD... It screamed along. OMF2097 ran smoothly. Broke a couple of Gravis Analog Pro joysticks playing that game... X-Wing... May have contributed to the death of one or two joysticks, too.

Warp Connect? Eh, wasn't all that different.

Warp 4.0? They messed with the GUI! Nooooooo!

eCS. Same GUI as 4.0, Support was... there... now and then... more than can be said about what IBM ever did. Never got Wifi to work on the portable I installed it on.

Haven't tried ArcaOS, yet. and my old Pentium is stored away right now.

[-]

Web-Lackey@reddit

Literally (Microsoft) LAN Manager? By the time of OS/2 2.1, Microsoft was long gone from OS/2 2.x, and IBM had released LAN Server as the standalone package that replaced OS/2 EE 1.x.

I‘m not saying you’re wrong: the version and timeframe is correct, and in fact MS was forced to keep selling OS/2 because it’s the only platform they had for running LAN Manager (and SQL Server, for that matter) until Windows NT in 1993. But MS LAN Manager ran on MS OS/2 1.3, not 2.x — which is why MS could sell it! I would guess it’s possible to run LAN Manager on OS/2 2.1…. I would just have to ask… Why?!? :)

[-]

Gadgetman_1@reddit

This is what 30 years of brain rot does to people...

Most of the servers were actually 1.3, running LanMan 2.2. We had a couple of special servers(would probably be better to call them workstations. Network supervision and stuff like that) that ran 2.1. Clients ran DOS 5 or 6 and the Lan Manager 2.1 client. (None of us ever dared to try take it up to 2.2) and this didn't really change until we switched to WinNT4.0 on clients.

It was a bit of a mess, with custom SW messing about in the logon scripts, menu systems on the PCs and so on. It worked...

Learned a lot about memory optimisation back then. Could get 639KB free in stanard memory WITH LanMan2.1(using NetBeui... Yeah, sorry) and a 40KB TCP/IP stack(Wollogong Pathway, from Australia, I believe) for when they needed Terminal emulators. The CADders really loved me.

[-]

TMWNN@reddit

Going to winNT 3.51 was a huge downer. Literally. Uptime was NOT included in that package...

Odd; I thought 3.51 was and is well-regarded for stability.

[-]

Gadgetman_1@reddit

Compared to OS/2, WinNT3.51 was a leaky rowboat about to hit the Niagara falls.

[-]

OldsMan_@reddit

I was using OS2/Warp back in the time . If I remember well I had 486SX 25 . Really loved it, all my colleagues was fighting with windows 3.0 or 3.1, I was the only one who had no freezing PC all day :)

Top of this I was developing in ASM , so when something went wrong just closed the frozen window. OS2 handled very well .

[-]

Martipar@reddit

That's because a ton of stuff didn't run on it so it didn't have rogue programs making it unstable.

[-]

BCProgramming@reddit

OS/2 could run Windows programs though and when they crashed the stability of the OS wouldn't be affected the way Early Windows versions are.

[-]

Martipar@reddit

OS/2 couldn't run all Windows programs.

[-]

BCProgramming@reddit

The only quirk I'm aware of is that the compatibility was effectively "Standard Mode" Windows, but I'm not sure if there were many applications that actually required 386 enhanced mode to function.

[-]

OldsMan_@reddit

Even Solitaire could crash windows 3.x :)

Seriously: win 3.x was lightyears behind os/2 .

[-]

Martipar@reddit

That's a gross exaggeration.

[-]

-jp-@reddit

Not really. I’ve never seen Solitaire specifically crash Windows, but there’s no reason any similarly trivial program couldn’t. There was nothing in the way of memory protection; nothing forcing a program to yield control to let other programs run.

Mind you, I’m not dumping on Windows specifically. It’s just a consequence of its design. I remember Classic MacOS had a similar problem, where just opening a menu hung the OS and every app because it didn’t give up control until you selected an option.

[-]

agent_flounder@reddit

The thing about menus sounds right. Prior to MacOS X, multitasking was cooperative vs preemptive. Your application had to call a function periodically to release control to other apps.

[-]

sirflatpipe@reddit

Windows 3.x was cooperatively multitasked (at least in the Windows VM) and without memory protection, any badly designed app could take down everything.

[-]

kweiske@reddit

It was pretty ambitious. I ran os2 2.1 on an IBM PS2 model 80. 386 dx25, 8 mags of ram, 70 Meg ESDI hard disk.

We were able to run Microsoft word in Excel for os2, connect to a Microsoft lan manager Network, Annabelle network, connect to as400, and connect via dial-up to my PBX. Windows 3.0 at the time would have burst into flame trying to do all that.

Thankfully for windows, os2 was a lot of work to keep running. There's a lot of configuration file tweaking to get it working right. Windows 3.0 crash a lot, but you could restart quickly.

[-]

Consistent_Cat7541@reddit

OS/2 did not take off as a consumer OS, but had good penetration in corporate environments. If you want some interesting insights on how IBM mishandled it, see In Search of Stupidity, by Merrill Chapman.

[-]

khooke@reddit

I worked in an office in the early 2000s where was a PC sitting in the corner of a storage room. It was the voicemail system for the company running on OS/2

[-]

Psy1@reddit

I don't think IBM really planned OS/2 to be a consumer OS more a competitor to Unix that was taking off at the time. Yet given IBM's PS/2 strategy OS/2 was also expected to fit that role and again with IBM's PowerPC strategy that gave us a rushed PowerPC port of OS/2 Warp.

[-]

Der_Unbequeme@reddit

I installed OS/2 Warp4 in VirtualBox, and it runs slower and worse than Win11 21H2, also in VirtualBox, in the same environment.

[-]

marmarama@reddit

OS/2 does some pretty esoteric things that were quite specific to real 286/386/486 CPUs, and modern x86 hardware virtualization doesn't virtualize.

As such, you're lucky if it runs at all as a VM. Where it does, it's usually using hybrid virtualization/emulation where the esoteric functionality is trapped and emulated.

I have the best results running it in full emulation with JIT, such as in 86Box, on emulated hardware that is contemporary to the version of OS/2. But yeah, that's slow.

[-]

scubascratch@reddit

Could you go into any more detail about what OS/2 was doing on these CPUs that is different than other OS?

[-]

itsasnowconemachine@reddit

https://www.os2museum.com/wp/why-os2-is-hard-to-virtualize/

[-]

-jp-@reddit

That doesn’t surprise me. VMs rely on drivers to perform well, and ofc they are going to prioritize the current version of Windows over a 30-yo version of a niche OS.

[-]

mnlx@reddit

It runs great in Virtual PC! If you keep a Windows XP box, even with a PIII it's an interesting virtualization option. I'm not masochistic enough to install it in actual hardware.

[-]

khooke@reddit

On real hardware period to the time it runs great. Also, ArcaOS (OS/2 updated to run on current hardware) runs great either virtualized or on real current hardware.

[-]

MonkeySmart409@reddit

Everything was better than windows - It wasn't until windows NT / 2000 / xp that microsoft made it into the modern world of protected mode ushered in with the 386.

I was had my nearly crash proof OS running crazy stuff, executing on my friends machine and rendering on mine remotely via xwindows, total OS awareness of the internet, protected mode that decoupled program crashes from the OS, a full suite of free compilers and other stuff microsoft was charging you for... all on linux when the rest of the world was still looking at windows 95 blue screens of death because solitaire had a whoopsie.

Back then I knew bill gates had made an unconscionable negative impact ok computing with the anticompetitive practices he used to put windows on top.

Now we know he's also a pedo. He's thoroughly a piece of shit.

[-]

sirflatpipe@reddit

Windows got 32-bit support with the introduction of Windows/386 in 1987.

[-]

MonkeySmart409@reddit

When it "Got 32-bit support" doesn't have anything to do with my comment - which was about protected mode and other capabilities.

In windows/386, apps and the os all shared the address space.

The user/kernel separation of NT, unix, etc was not there - except for running dos shells in virtual 8086 mode. Full blown Windows apps had no separation and still relied on 'cooperative' multitasking.

And I interacted with hundreds if not thousands of PCs in this era and never once encountered Windows/386 outside of magazine articles.

[-]

-jp-@reddit

The hostile tone seems entirely unjustified.

[-]

HandGrindMonkey@reddit

The first truly multi-tasking OS, so yes. Not many native applications, and until the advent of Warp (the first usable GUI IMO), it was stuck in a specialist niche.

I particularly liked Warp, I fairly certain this prompted Microslop to up their game.

[-]

OS/2 1.3 was a lot more like Windows 3.0 in terms of look, feel, and functionality.

[-]

DrumsKing@reddit

Windows NT for the win.