25 years of OS/2: the legendary failure that refuses to die. | Hacker News Reader

25 years of OS/2: the legendary failure that refuses to die.

techland.time.com

136 pointstechnologizer61 comments

I tried to give OS/2 a try, but what happened was kind of bizarre. I had a DEC PClone with a 486 (cannot remember if it was DX or SX) inside. OS/2 needed a graphic driver for the S3 chip. I called DEC support to ask for the driver and they said they didn't have it but IBM did. The DEC rep kept me on the phone while he called the OS/2 support number. DEC had incredible service at the time.

Then the bizarre happened. The IBM support person said I needed to sign an NDA to get the driver. Both the DEC rep and I tried to explain to him I was a humble end user and not interested in anything but the driver. There must be a mistake as I did not want source code. Nope, just the driver required a NDA and some verification. I said I would think about it, and we hung up. The DEC rep apparently had quite a few people gathered around him and they were laughing pretty hard. He then asked if I would like a nice copy of Win NT with no NDA and all the drivers for my machine.

I really wanted to like OS/2.

I used to do OS/2 support as an IBM co-op back when I was in college. In many ways it was pretty sweet -- I did some cool stuff with the built in Rexx interpreter, which -- though not as or widely supported as, say, perl -- was still pretty nice.

Regarding your specific problem, I'd add that IBM's support structure was byzantine, with multiple groups with different agendas and tons of snafus like: "this isn't our issue, contact Thinkpad support," followed by "this isn't a Thinkpad issue, contact OS/2 support." (Aside: I once tried to get a different IBM group to handle an issue I believed to be theirs, and after getting rebuffed multiple times, I wrote a rather angry response which got me in trouble with my manager. Learned a lot from that -- mostly, don't write angry.)

9 times out of ten, the issues we received boiled down to a) missing / out of date drivers for some piece of hardware (as in the S3 chipset you mention), or b) incompatibilities/errors running existing windows apps. I think OS/2 failed for a variety of reasons, but ultimately, even though it was better than windows (for a while), people just didn't feel a strong need for pre-emptive multi-tasking, so it was never a strong selling point, especially if it meant you couldn't run the latest PC games.

I realize this is off topic, so please forgive the following rant. I'm not sure if it's out of principle or just stubbornness, but it's gotten to the point where every time I read an article that is paginated (most likely only because whoever runs the website wants more page views), I refuse to continue to the next page and end up reading only a fraction of the article.

My attitude towards pagination wasn't always this way. Over time, however, it's just annoyed me more and more. When I get to the bottom of a page as I'm reading an article and see that I'm going to have to click on these tiny numbers multiple times, wait for a new page to load each time, all while interrupting my train of thought, I get so annoyed that I immediately say "NOPE!" and hit the back button, no matter how good the article is.

And out of principle, I'm not installing any 3rd party software (like readability, etc.) to handle this for me. I shouldn't have to. Site owners who run articles need to find better ways to get page views, ways that aren't inconveniencing everyone using their site. Some might say "well that's just how it is!"... and I say only if you allow it to be.

So as a bit of a followup to this rant, if you submit articles like this to HN and there's a printer friendly version (or even a "view all" version), please submit that URL instead.

I spent a few minute trying to find an "all pages" version of this article and failed. Even the damned print link just prints one segment!

I have the same feeling, the most annoying thing is that i expect a specific amount of text.

When i see the scrollbar, i think this will take 2 minutes and the topic OS/2 is worth 2 minutes of time, but not 20 minutes. I feel cheated because i don't want to read it all and the first part was somewhat useless because i didn't get the whole story.

Readability didn't work for me on this page anyway. Very frustrating that content producers do this and provide no way to view the whole article in a single click.

As far as first world problems go, this one is really annoying. Especially if you send an article to the Kindle and don't notice that you only got a part of it. After a few times, I learnt to view and check first but it really kills the flow.

I have to admit... I did have a "First world problem" meme image in my head as I posted this and thought it might not be an appropriate complaint to raise on HN.

First world problems. :)

PS: This is only as off-topic as your reply.

Someone has to produce those interesting articles you consume for free, but he probably doesn't want to work for free. Would you?

Of course I would.

Reading the newspaper must be no fun for you then.

I spent a little bit of time trying to become an OS/2 fanboy. Remember that this was before Windows NT (and Linux was a gleam in Linus's eye) so a PC that could walk and chew gum at the same time seemed like a miraculous feat for anyone used to DOS. It was a pretty sweet development platform for the time, since you could run multiple DOS and Windows 3.x sessions -- I even set up two Windows sessions to communicate over null modem to test a game.

However, there were defects both large and small that didn't seem to get resolved -- from missing files on the custom Windows installation to filesystem corruption when the swap file filled up (this last one was the final straw for me).

Still it's interesting to think how its shared DNA is in Windows today, just as NeXTStep is in Apple's products. In fact many APIs between Win32 and OS/2 are similar (I once made some OS/2 API bindings for Borland Pascal).

What version?

I think I gave up before Warp so only OS/2 2.x -- the API bindings were for BP7 I think. There was a German computer magazine that released a patch for the compiler to output OS/2 executables.

I remember buying multiple copies of OS/2, but never installing it, at one point it was so discounted that it was the cheapest way to buy good quality floppies. IIRC warp came on about 40

Yes indeed. I remember thinking I had gone to heaven when I got my hands on a Warp 3.0 CDROM. Installing any OS that has 40 floppies is painful. Kind of reminds of of the slackware days as well...

Wow, this brings back memories. I was a total OS/2-Head until about 1993 or 1994 when I got both an Yggdrasil Linux CD for my 486, and, a NeXTCube; and from there, ended up a Unix fan.

The problem with installation, was that all development was done on PS/2s. The developers never did much testing on non-MCA (MicroChannel Architecture, a pre-PCI intelligent bus design) hardware.

That is why, for instance, when installing OS.2 2.0 and 2.1 on certain PCs you had to disable the CPUs L2 cache by using the legacy "non-turbo" mode and do part of the installation in 8Mhz AT mode; otherwise the installer would crash and you wouldn't know why.

I think a lot of OS/2 users went to Linux and their disgust for MS' admitted astro-turfing and plain outright lying, fueled Linux adoption.

ADD: OS/2 ran well in 32MB and flew in anything more than that. Unfortunately at that time, 32MB was hundreds of dollars, and some systems couldn't even take that much RAM.

Hey, nice to meet you .. I also had Yggdrasil (and SLS!) systems back in the days of OS/2, which I very briefly flirted with (IBM compilers, yay!) then seguéd back to MIPS and RISC/OS and Irix and SGI (grr.. why you no make laptop, SGI?!! wtf..) with a bit of a dodgy diversion from Linux in the 90's through NextstepX86, sustained a lusty affair with a BeBox, until .. like a lot of other people .. here I am with my 'best unix workstation ever', an Apple.

Apropos the living-on factor of OS/2 nuts keeping it running, its a good, healthy thing to 'not-abandon' technology.

There is a scale to this.

I personally really like the guys at forum.Defence-force.org, burning the machines in current glory with new code. The Oric-1/Atmos is still a computer, and its values during its market period rose (and fell!), but yet the working machines, today, are still getting new code written for them.

It is actually some sort of utter joy to have new titles for a very old, memorable machine or system, which can still be enjoyed!

Computers don't die. Their users do. Lets hope the kids can still turn every single one, on ..

OS/2 was actually quite good on underpowered hardware if you stuck to console apps running fullscreen. I started with OS/2 2.1 on a 40Mhz 386 with just 8Mb RAM, connecting to BBSes and reading FidoNet with a newsreader called Blue Wave and programming with the Boxer editor. OS/2 Warp was the first shrinkwrapped software I bought, still on floppy disks; I was quite proud of that as a 14 year old, as piracy was rampant in Brazil at the time.

I remember OS/2 was quite popular with both BBS operators and BBS users around here, as it was the easiest way to multitask on a PC in the early 90s. It had a quite vibrant shareware/freeware scene. You did have to be picky about hardware, though, but even here it was easy to find video cards, sound cards and modems that ran just fine.

I really liked OS/2. I could run several varieties of DOS under it and early Windows - very flexible when I needed to develop using several operating systems.

The real reason OS/2 was not widely adopted? Maybe this: the first time OS/2 was booted after an install, the bootup time took a very, very long time. I was once in a computer store and some guy was practically screaming at the sales people. He had bought OS/2, installed it, and the next time he used it hit the one time only long bootup time. He was ranting how he immediately took it off and reinstalled Windows. There must have been at least 10 other customers who heard this.

I wonder how often this scenario happened.

Huh, funny. I always think it's interesting how boot times have increased over the years regardless of Moore's law. My Apple ][ took about 6 seconds to boot, yet a modern PC or Mac takes much longer (and even longer to shutdown, usually)

My Sinclair ZX Spectrum clone, on an 8 bit CPU at 3.5 MHz, 64 kB total memory, used to "boot" in a second or so.

I don't think it was because of the quality of OS/2 itself. It's always the applications that make or break a platform. I remember trying OS/2 Warp and the OS itself did impress me. But most of my favorite games and applications did not run on it. After a while there was nothing that kept my interest so i went back to Windows.

Never used OS/2, so maybe you can clarify this for me:

I thought that one of the limitations of OS/2 was that while it could run DOS programs, it could only run one DOS program at a time (in addition to a number of native ones). Is that correct?

I don't think that's correct but can't remember well enough to say so definitively. You could certainly have multiple DOS and Windows programs "running" but whether they were all active simultaneously or not I can't remember.

No, OS/2 2.0 and later could run any number of DOS and Windows 3.1 applications at once. I'm not sure about the 1.x series. You're probably remembering Windows 9x's feature of being able to automatically reboot into MS-DOS for running games.

You are correct.

On an 80286, OS/2 could only run a single instance of DOS due to that processor's lack of a true virtual 8086 mode.

I wonder how often something similar to this scenario has happened when new Linux users have to wait a long time for their system to finish fscking on every 30 reboots, and asking how to skip it or keep it from happening gets responses along the lines of "ura luser", "lol you turn off your computer?", and "xyz has been banned from #linux (go back to Windows)".

Probably not that often. Usually it tells you to press a key to skip it. If they bothered to read that it's fscking then they would know how to skip it in the majority of cases. (I know that GNOME and XFCE do this. Pretty sure KDE does too when I used it.)

In the case of OS/2 This sounds like a problem that could have been solved with a well placed printf call during the boot routine. (Like say; at the the beginning.)

OS/2 was nowhere close to being as stable as Linux. There were all kinds of normal usage scenarios which cause the system to lock or crash. (OS/2 had the same "49.7 day bug" as Windows and nobody noticed for years.)

It is similar to how now a number of websites and individuals measure Windows boot performance on first/second boot, without giving various internal windows boot optimization procedures like Prefetch and ReadyBoot time to kick in.

I used to be the Editor In Chief of the OS/2 e-Zine! (http://www.os2ezine.com), although I began as just a contributor. The e-Zine was started by Trevor Smith, who slowly handed over more responsibilities until eventually retiring to do other things.

I have fond memories of that era, and of constantly working to keep your chin up as IBM's support grew more and more reluctant. Probably the most important lesson I walked away with was understanding how deep the roots of zealotry can go, how something as cosmically unimportant as an operating system can get woven into a person's identity until the two are indistinguishable.

I was at IBM in the mid-90s and OS/2 was my first prolonged exposure to GNU tools. I got in deep with TCP/IP as part of practicing to become a "Certified Engineer." We ran NCSA httpd and cgi scripts on it. Mucking with OS/2 and its configurability on a PS/2 tower (that you could about crawl inside of) lead directly to mucking with Slackware and working our S. American teams who'd created an MCA patch for the earlier Linux kernels. (Getting DOOM to run on that PS/2 386 was a banner evening.)

Marketing was always an uphill battle. Fans didn't like the Betamax analogy, but it was apt (w/r/t stronger specs and weaker sales). MSFT already knew how to market a dung pile.

I didn't realize NYC Transit used it. I still see it in ATMs and know UPS had a huge install, enmeshed base.

A few years back (around 2008, maybe) I stumbled upon an ATM that had failed to boot all the way and was pretty surprised that it was running OS/2 (and a Rexx script). This was for Brazil's largest bank (Banco do Brasil). Since then I think they've moved to Linux.

Same thing here, also 2008, but with an Allied Irish Bank ATM in Dublin. The machine crashed as I was using it and permanently relieved me of my bank card, but the up side was that I got to watch the entire OS/2 Warp boot sequence play out.

They made the move around 2008 or so.

But most ATMs move to Windows XP, strangely enough.

To watch them boot, you only need a power failure. Strangely enough, ATM is often not even plugged in UPS.

About six years ago, I was in a council bus terminal and the system running the timetable monitors had crashed. It was showing a Commodore64 prompt.

OS/2 Warp has a nice system. Too bad IBM refused to support it.

They had a golden opportunity, when it was clear that PCs would be transitioning from 16-bit operating systems to 32-bit operating systems. Microsoft had WinNT for 32-bit, Win3.x for 16-bit. WinNT didn't do a good job at the time running people's existing 16-bit Win3.x applications, so migrating to NT was a tough sell. This put MS in a hard place, because they needed something that ran existing Win3.x stuff reasonably well but supported 32-bit for the future--in other words they needed a transition system.

That transition system was to be Win95, but it was months away when Warp came out.

Warp had excellent Win3.x compatibility. In fact, it arguably ran Win3.x applications better than Win3.x ran them. And it had excellent 32-bit support.

IBM should have aggressively pushed Warp as the 32-bit successor to Win3.x. They could argue that instead of waiting for Win95, switch now and you'll get the benefit of 32-bit, and you'll get a better 3.x experience than you have now (or that you'll have under Win95).

What would Microsoft be able to do? If they countered by downplaying the importance of 32-bit, they'd just make people reluctant to switch to Win95 when that came out.

Instead, IBM basically just made Warp available, and then ignored it. Here are some examples of how they ignored it.

1. They didn't spend any money promoting it in stores. A lot of consumers do not realize this, but in retail stores shelf positioning is for sale. When you see a product featured in an end cap, or on a shelf at eye level, or in a high traffic location, it is because the manufacturer of that item paid to have it put there. If the manufacturer doesn't pay for good placement, their product goes on in inconvenient shelf in a less trafficked part of the store.

Guess where OS/2 was in Egghead and CompUSA? Someplace in back, where the lighting was poor, sitting on the bottom of a shelf.

2. They didn't give a fuck about existing developers. I recall reading a column by, I think, Jerry Pournelle, where he went to a trade show. First he went to the IBM booth, told them he'd heard about OS/2, and would like to develop for it. He asked what he needed to do.

They have him a form to fill out. It was an application to apply for developer status. It asked all kinds of details about what he planned to develop, his business plan for it, and so on, and there was an application fee.

Then he went to Microsoft's booth, and said he'd heard about Win95, and asked what he had to do in order to develop for it.

They gave him the SDK and tools right there.

3. They didn't give a fuck about beginning developers. Flash forward to after Win95 was released. Microsoft made developer tools available IN THE RETAIL MARKET. They were for sale at places like Egghead.

Think about that. Some kid playing with Win95 gets an idea for a game, and wants to start programming. He could go out to his local software store, and get everything he needs to start being a Win95 programmer!

Could you do the same for OS/2? Well, actually you could--but not from IBM. Watcom C/C++ was available in retail shops and could build OS/2 applications, and even had some licensed libraries from IBM included. However, Watcom really was aimed at the professional developer. The licensed documentation wasn't as complete as the Microsoft documentation for Win95.

4. They didn't give a fuck about developers who actually developed for OS/2. If you persevered, and actually made OS/2 software, you were on your own. They wouldn't help promote your application. They wouldn't feature it in any ads or provide any co-marketing funds to help you promote it.

An IBMer once shared a joke with me:

"Q: How do you rid the world of illegal drugs?

"A: Legalize them and put IBM in charge of marketing."

I can relate as a one-time business partner. Our biggest competitor was IBM Global Services. Once we even had a teleconference to discuss going after an opportunity. We dialed in, but no IBM. After waiting for 15 minutes, we called someone who was to be on the call. "Oh... we decided to do it ourselves." Didn't even bother to tell us.

Didn't Microsoft win by putting itself into position to preinstall its OS on new computers?

Preinstalls discourage other operating systems for two reasons:

1. they set a functionality mark that a on-preinstall OS needs to surpass in order to interest someone in switching (especially if the new OS costs money),

2. they enable inaction. What I mean by this is that you may find the features of a new OS interesting enough to want to try it, but doing so actually will take some effort. If you procrastinate, you've still got the old OS.

Before Win95, the preinstalls were generally either DOS, or DOS + Win3.

When Win95 came out, it was an immediate hit via retail sales. On launch day, many stores were open at midnight to sell it as soon as they could, and people lined up to buy it. I don't have the numbers handy, but I'd be surprised if Win95 on the first day of availability did not surpass OS/2.

This shows that there was a significant number of people in August 1995 for whom an OS of Win95's level could entice them to want to try it sufficiently to get them to actually go out and buy it and install it.

If IBM had properly marketed OS/2, those people could have been their customers. IBM could have established themselves as the preferred 32-bit PC OS before Win95 got out of the gate, and vendors would have been rushing to sell preinstalled OS/2 systems.

One can make a good case that after Win95 came out and was getting preinstalls, no amount of effort by IBM could have overcome that--Win95 was good enough that OS/2's advantages probably were not enough to overcome procrastination. But by the time Win95 was getting preinstalls, Microsoft had already won from IBM's indifference pre-Win95.

Related to points 2 and 3, Microsoft didn't just make it easier to get the Windows tools, they made a tool that made it easy to write Windows programs: Visual Basic. Everyone loves to hate VB, but it was a huge factor in enabling line of business developers -- people who just need to bash out CRUD apps as quickly as possible, people who don't need the fine control of C/C++, people who want to knock together a bunch of forms rather than mastering the vast Windows or Presentation Manager APIs -- to create applications on Windows. And of course VB was also adopted by hobbyists and people creating simple utility programs, again ensuring that those programs would be available on Windows but not OS/2.

I was an OS/2 user from 1993 - 1996 or so. I have very fond memories of it - for those lucky enough to have a supported machine, it felt indestructable. You could do the most obnoxious things inside a DOS session, and the most that could be trashed was the DOS session itself. At the time, I was doing x86 assembly and C programming for a college course. The kind of pointer mistakes that would cause (or necessitate) a reboot on DOS only took out the session on OS/2.

I only stopped using it because I changed computer to one that didn't have a nicely-supported Cirrus video card (my replacement machine had a Diamond Stealth), and there were no drivers for the new machine. Graphics drivers on OS/2 are a separate story! Two years later, I was on Linux full-time, but I still find myself longing for a Workplace Shell clone.

OS/2's big problem was it needed 8MB of RAM to run acceptably. 4MB (the stated minimum requirement) was just enough to boot it up and, maybe, run one very light weight application (with the HD thrashing trying to swap memory in and out). Things where worse with Win-OS/2, which was a full copy of Windows 3.1 running in a VM. You really needed 16MB to take true advantage of OS/2 multitasking capabilities. IBM should have billed it as a workstation OS until consumer hardware caught up, just as Microsoft did with Windows NT.

That's exactly why Windows 95 succeeded. At the time 4mb of RAM was a lot. But pure 32-bit OSes like NT and OS/2 ran 16-bit apps in their own full VMs. Which is great from a design perspective but lousy for end-user performance on average hardware since it meant about 4mb of overhead per app. Win95 ran all 16-bit apps in a shared VM and used a few other quick and dirty tricks to keep RAM usage low even if you were running lots of 16-bit apps. The end result was a system that traded maintenance cost and stability for better support for older apps on more modest hardware. In the end it turned out to be one of the most successful software products in history, even though it gave MS so much technical debt it took them nearly a decade to work through it.

My first computer had OS/2 Warp 4 installed and I loved it. I remember checking out the first SuSE Linux release at that time and I was puzzled how primitive it looked compared to what I was used to from OS/2. A few years later, I switched from OS/2 to Linux and have been using it ever since.

1.44" huh? Not 3.5" 1.44MB maybe?

This is why it never saw mass adoption. Strange, tiny, proprietary distribution method that nobody could use because no companies bothered to make a 1.44" floppy drive. You had to build one from scratch if you wanted to use OS/2. (The best place to start was by harvesting the motor from a larger drive.) Thus, only the true tinkerers and hackers ever made it past "Step 1 - Preparing Your System."

(Somebody went and edited the main article, thus rendering this thread moot. Quite a shame, really. Typos can be fun. ;) )

I remember my dad having a Zenith MinisPort with a 2" floppy drive. Even as a kid I remember thinking "Ok these are just ridiculously small."

The OP is also the author of the article, so I'm not surprised he went and fixed it. But I'm a little miffed he didn't even give me a thank you.

We stumbled upon either a 2" or 2.5" floppy early in my college career. None of us had ever heard of such a beast - it was a bit like alien tech landing in our laps. We figured it was "the future" and that smaller floppies would see mass-adoption, as some sort of CD-killer.

Then our professor destroyed our dream world and said they'd died off in the 80s. (And then wisely predicted flash-like memory would kill off all of them.)

There's a wide range of weird floppy sizes. Perhaps one of the best known was sold for the Amstrad PCW - a popular word processing system. (This machine kept CP/M alive during the 80s and into the early 90s.)

(http://en.wikipedia.org/wiki/Amstrad_PCW).

There were others. Here's a 2" drive - yet another failed Sony media format. (What's their ratio of successful : failed formats?)

(http://en.wikipedia.org/wiki/Video_Floppy)

OS/2 was one of the first CDROMs I ever owned, when 1x drives were on the market.

The title is a bit misleading. OS/2 (especially Warp) wasn't a failure from a technical standpoint. Rather, IBM treated it like a redheaded stepchild which detracted from its core systems of mainframes and AS/400s, because--you know--this pc on every desktop thing is just hype.

As an aside, this reminded me of an old Dilbert from that timeframe (mid-1995): http://dilbert.com/dyn/str_strip/000000000/00000000/0000000/...

OS/2 is still in use in NYC's subway MetroCard system... shocking. The city is probably being squeezed out of tons of money to maintain technology that is ancient.

I don't find it all that surprising. A few large insurance companies still have a lot of their (core) software written in COBOL and outsource its upkeep to IT companies. It's a big expense, nobody likes it; but replacing it would cost more.

I learned multi-threaded programming on OS/2 (coming from VMS) and it was a pleasant experience.

Microsoft's OS/2 Crush Plan makes a good read today.

http://techrights.org/wiki/index.php/OS/2_Crush_Plan

Also see http://news.ycombinator.com/item?id=3785819 . I posted a lot of comments to it.

Reply on news.ycombinator.com