Where were the bullet holes on OS/2's corpse? Its head ... or foot? My last piece on OS/2 was in part a mea culpa, a history of my part in its downfall. However, I can't claim all the credit. In fact, if I'm honest, there were hundreds of reasons why OS/2 failed, and most of them had nothing to do with me. So, here are some of the real corkers. Once upon a time, IBM made extraordinary money …
re: OS9. Yup, Microware would have had something to say.
Yes, kids, before MacOS had its version 9, there was an OS9 that wasn't anything to do with Microsoft, nor Apple, nor even IBM.
'Course the original version required a 6809-based machine, but at the time I had one of those.
(Showing my age again, I guess.)
My first experience with OS/2 was that Microsoft shipped a copy of it (I think version 1.2) in the box with SQL Server 1.1 (a partially implemented port of v4.2 of the Sybase product). This was because until NT they did not have a version of Windows that could actually run SQL Server (or any real server product).
My other memory is the two enormous stacks of floppies (20 or so each I think) that I had to work through in order to install OS/2 2.0 and the first service pack on a machine (I think to run the old Lotus Notes Server).
So what it boils down to is that OS/2 was more advanced than the competition but too expensive, it needed more resources and was only designed to work with IBM hardware - any other platforms' success being just a lucky co-incidence.
Put aside all the infighting and cluelessness about marketing, internationalisation (where I was working during this time, none of the "internationalisation team" even had a passport) and the turf wars - every new product has that. It's just lucky they didn;t have software patents to worry about.
Given IBMs background and way of doing business at the time, I can't see how such a foray into the commercial world and with a non-IBM partner could possibly have worked out any differently.
> So what it boils down to is that OS/2 was more advanced than the competition but too expensive
Pretty much. I remember when Win XP came out everyone crowing about it being a truly 32-bit OS. Yet I'd been using a truly 32-bit OS before even Win 95.
It's something that happens over and over again. I've heard people crowing about C# and Winforms and saying how much better it is than C++ and MFC because it was RAD. Some of us were doing C++/RAD development with Borland builder in the late 90s. Then there's the Betamax/VHS/v2000 sequence.
What it comes down to is that in most cases marketing trumps engineering. This is a lesson everyone reading El Reg should keep in mind. It can grow or wreck your career :)
Same here .. I never managed to convince myself to shell out the money to buy OS/2.
I wanted to get Warp, becuse it could run windows better then windows itself , but I never could convince myself to spend the money on it.
As for running 32bit, back then I was doing most of my development on SunOs (4.X) or Convex (with some dabbling on IBM AIX 3 and even a Dec Alpha (cannot remember what the model was), so 32bit was so common for us, that we failed to realize why MS was making so much noise about it :)
OS/2 was actually price competitive by the time of OS/2 Warp, especially if you bought the Red version without Windows in it. Windows didn't catch up until NT 4.0.
However, Windows 95 had two non-technical advantages over OS/2 - it looked nice and it behaved sanely.
It also enjoyed one technical advantage which used to piss me off no end with OS/2 - every process had its own windows message queue. In OS/2 if one process was in its WNDPROC, then every other process was blocked. If the process never returned from its WNDPROC (e.g. infinite loop, debugger breakpoint etc.) then the whole GUI hung.
Anyway IBM's failure was 50% their own internal politics and lack of vision, and 50% Microsoft sticking the boot in, engaging in anticompetitive practices such as licensing deals where PCs paid a Windows tax whether they shipped with them or not. I remember even IBM PCs used to ship with Windows. When a company can't even persuade it's own PC wing to ship an OS, then they may as well give up right there.
> every process had its own windows message queue.
We used ccMail for a time back then. It was actually quite a nice idea to have it integrated into the WPS (using the OOP features I mentioned in a previous reply) but when it locked it would lock everything as you say. But I thought the separate message queues was actually introduced by Warp. I know that one of the last version I used had the separate queues and I didn't think I used anything after Warp. Hmmm. Wikipedia says there were different versions of Warp. Warp 3 rings a bell.
Meh. Fading away into foggy recollection as I age. Probably all for the best :)
"Warp 3 rings a bell."
OS/2 Warp was OS/2 3.0. To my recollection it didn't fix the message queue issue. I had to use a console command called watchcat to kill processes which hung. I'd stab some key combo which put watchcat to full screen, kill the offending process (normally the thing I was writing) and proceed. Still a pain in the arse.
I was done with OS/2 before Warp 4 (OS/2 4.0) so I didn't have any first hand experience.
It would not surprise me at all if the single message queue thing was easily fixable, but IBM being IBM chose not to for fear of breaking badly written software which used the message queue as some kind of primitive sync lock.
However, Windows 95 had two non-technical advantages over OS/2 - it looked nice and it behaved sanely
There was one other major advantage - software availability ... I kept with OS/2 all the way from 2.0 through to (I think) v4 (the one with vocie recognition built in as standard ... sadly as I bought a copy while on a business trip to the US it only worked if I adopted a faux US accent as it had the US speech patterns built int!). It was way ahead of windows in terms of internet connectivity at that time but when in early '98 I went to California to work on a joint project my employers we doing with a company there I relented on getting win 95 so that I could run a Hauppage TV card which would allow me to keep my then 2 year-old son happy with Teletubbies VHS tapes played rom a PAL standard VCR ... however I intended win95 to be only used for that and retain OS/2 as main OS. Then I went to Fry's electronics and saw several aisles of win95 programs and 1 shelf of OS/2 utilities (only ever once saw a copy of Stardock's Galactic Conquests which was the only real OS/2 game) and I realized that it was time to give up and switch to the "not the best, but good enough" win95.
N.b. while it may annoy the fanbois, I maintain that the iPhone success over recent years has been down to the same effect - it may be technically inferior to other options but the amount of apps available coutner weighs that. However, Android, unlike OS/2, seems to have a company behind it which is determined to make it a success
I gave up on OS/2 around the time that Stardock Systems (Brad Wardell) gave up on OS/2. By that time about the only people that used OS/2 were Germans - gods only know why.
Software availability was so bad that there wasn't even an offline usenet client for years - the only native ones available assumed you didn't pay for dial-up local calls. Mail clients were bloody terrible until PMMail. There was a decent graphics package but I can't for the life of me remember what it was.
The bottom line for consumer machines is software availability. If you looked hard enough and paid enough you could live with OS/2, but it was a PITA which soon became tiresome.
Very good OS, shame IBM was a crap company at the time.....
Software availability was so bad that there wasn't even an offline usenet client for years - the only native ones available assumed you didn't pay for dial-up local calls.
Changi (not a Usenet client in itself, but a local news spool thingie like leafnode)
Mail clients were bloody terrible until PMMail.
That's what kept me on OS/2 (eCS) until about two years ago.
There was a decent graphics package but I can't for the life of me remember what it was.
PMView.
I think it was Colorworks rather than PMView. Not 100% on that though.
For usenet I just ran a Win-OS/2 session and KA9Q, can't even remember what the client was - probably whatever the Demon package had. Some USAF guy stationed in the UK developed a decent native OS/2 client which worked on/offline around 1997 but I can't remember whether it ever got finished or not.
I have used OS/2 2.1, Warp 3 and Warp 4 and I feel that if IBM had given their bastard child a bit more love and attention, it would have really crapped all over Windows. But the lack of driver support and application support was the killer for me. I loved that it was a proper 32 bit operating system at the time when you either had to use a 32 bit windows shell over DOS 7.0 (Windows 95 & 98) or you had to go to Windows NT 3.5/4 for a proper 32 bit operating system. NT was good but didnt have the application support from Microsoft and the majority of other application developers were still farting about writing code for DOS and Windows 95.
In the end, Microsoft won that battle because they took all of their cool code that worked on the PC-Compatibles and put out Windows NT (Microsoft's version of OS/2 3.0). And with their access to the undocumented Windows API 32 bit subsystem, they had a lot more application support than IBM allowed.
It's a shame really because I really thought that OS/2 had the legs to really make it.
Never did get Warp running on my home brewed box. Even made sure all of the hardware was on the supported list. Had all of the necessary drivers. It just refused to work. I wasted a couple of weeks on it.
Having given up on Warp, I bought Win95 and it just worked straight out of the box. I sold my Warp disc to a friend who, eventually, got it working on his PC. Took him several weeks though, if memory serves.
Started with 2.11 on a 486/66, 8 Meg memory, ATI VLB video, Adaptec VLB SCSI card - no problem, except the tedium of shuffling those umpteen floppies.
Upgraded to 3.0, no problem (this time from CD, yay)
Moved to an AMD K5/450, 32 Meg memory, ATI PCI video, Adaptec PCI SCSI card - no problem again, just a bit of bother actually getting hold of a driver for an ISDN card, but it worked rightaway.
Upgraded to Warp 4 - once again, no problems.
Upgraded the system to an Athlon 1000, 512 Meg, moved the cards and disks - no problems
Upgraded to eCS 1.2 - no problems once more (actually, I think I did a reinstall that time; all my apps and data were on a different disk, so except for recreating the shortcuts no big deal)
I got Warp running on an Apricot which had 12MB RAM. The first installation attempt keeled over but gave a sensible error message. Posting that on CompuServe got me a very prompt answer which solved the problem.
But to be honest I found the colours ugly, and Windows 95 was just a couple of months away, so I lost interest.
I remember the marketing campaign for OS/2 Warp - on TV! A bunch of giggling kids watching one of their mates doing "something" supposedly on a keyboard (but they may have been watching alligators copulating for all we knew) and spouting "I didn't know you could do that on a PC".
Then installation support. 60 days. Which sounds a lot unless you have to try umpteen times to get a new, unfamiliar OS up and running while trying to do your regular job. When I decided to give it another final try and called support the first thing the frigging IBMer said after checking my record was basically "Your 60 days are up. Eff off" whereas it had been less than 60 minutes in total. I know that is how support works but didn't IBM realize they had a fight on their hands?
Ultimately though I seem to recall the main problem was the sales guys in the Personal Compuing division (which was hardware). They wanted to sell boxes, as many as possible, and didn't give a toss about which OS. It happened that they could sell more boxes with Windows on them. Talk about the enemy within.
I was running Warp in 8 megs on a 486. Still, if you wanted to do multimedia with full motion video you had to fork out a few quid for "big" hardware.
Note for the kidz: You could buy a second hand car for the cost of 8 megs of RAM in the early 90s. (Yes, I really did mean to type megs and not gigs.)
Where's the Four Yorkshiremen icon when you need one?
Note for the kidz: You could buy a second hand car for the cost of 8 megs of RAM in the early 90s. (Yes, I really did mean to type megs and not gigs.)
I can't remember what it cost for 8 megs for our work PCs, but I upgraded all 3 of them to 12 MB and both Windows 3.11 and Windows 95 ran sweetly in that.
It was some £350 for the upgrade from 4MB to 12MB on my 1995 Toshiba laptop.
At the same time a Vauxhall Cavalier in decent condition could be had for £500.
I can't remember what it cost for 8 megs for our work PCs, but I upgraded all 3 of them to 12 MB and both Windows 3.11 and Windows 95 ran sweetly in that.
I can remember upgrading my PC to the unimaginably massive amount of 16MB or RAM ... I bought 4 4MB sticks of RAM second hand via something like uk.comp.forsale ..... and I paid £320 for it which at the time seemed like a bargain. Earlier this year in an upgrade I bought 8GB new for something like £30 (though then again just over 30 years ago I remember great excitement when someone at school put together a 4kB RAM *card* so that we could play the text "star trek" game on a 6800 SWTPC system!)
Oh yes. I had a massive 4 megs of RAM in what looked suspiciously like an IBM AT (although the fact that it RAM checked in seconds and then paused while the selection of SCSI disks inside span up sequentially[1] was a bit of a giveaway) next to my desk.
Oh the fun you can have when you do all the PC R&M for a company and the budget for same is yours to control.
[1] 'Cos the card could do it, it sounded cool and starting 'em all simultaneously ran the very real risk of blowing the original, honest-to-god AT PSU. That's why.
.. as I recall, in the 80s, IBM was most famous as the World's Biggest Plumber on account of all of it's big iron being water cooled.
As well as heating universities, their other big boast was that there kit was 'proven', i.e. YEARS behind everyone else's. Not that there were many everyones out there: IBM essentiall was the computing market. The next biggest company - DEC - were absolutely tiny by comparison. And also tried to do a Windows-y O/S around the same time as OS2 - DECWindows - which also sank without a trace. Along with the company.
Err, DECWindows was an implementation of the OSF windowing system (yet another open standards initiative wot sank) for workstations - it competed with Motif; it certainly had nothing to do with MS Windows or the PC market.
DEC went through a number of attempts at an in-house PC, from the fairly neat (I seem to remember a dual-processor box, maybe Z80 + 6809 ?) to the unbelievably messed up (the Professional 350, because everyone wants a hobbled PDP-11 on their desk).
OS/2 Was hyped as the new thing that was going to send Windoze to a well deserved grave. It didn't require DOS, it could address a decent amount of memory, It was going to be much faster. It was going to be a true graphical OS, not a shell stuck to a command prompt. It promised the arrival of new and great native versions of WP, 123 and Symphony, Paradox and so on.
When we finally got our hads on it, we had built custom machines for it. 8 megs of ram, a 200 mb HD, grahivs cards with 256K or 512K ram. They were so hot the guys supplying the parts thought we were building the starship Enterprise.
Then came the disappointment. It was horrid to install, crashed every time you went for a pee, and gave you grief if you wanted to install DOS software, which you had to, because the great new native apps never arrived.
After several months of trying to make these boxes work anywhere near reliable enough so we could demo them to clients we simply gave up and returned to MS.
For me, OS2 was the biggest let-down in computer history. I still think it's a shame because it had serious potential. But I guess IBM never saw it as something they could flog to PC users as a product. Triple shame.
"IBM made more money out of its database sales than the whole PC division put together, and it vetoed any attempt to bring out a database that was more user friendly than a cornered rat on a house of cards."
So saddening. My immediate thought was, "This is what Lotus Approach is goung through, buried under Domino and Notes, and AS####,"
What a waste. IBM shows it has not learned thst CUSTOMERS make the company. I wrote in the Pt I story what i felt about the wish to see IBM release Lotus Smart Suite,. But, there will always be trotted out the excuse, "We cannot locate all the affected shareholders and royalty-eligible people..."
They had since 1997 to prep letters of future release and options and could have advertised to every tech and consultancy over the Internet that "If you are a shareholder, investor, licensee, licensor, co-inventor of anything in Smartsuite, you have 10 years to make yourself known so we can carve out a reasonable and equitable payment or new contract with you... This "Linux thing" might take off in the future, and Smart Suite might be our springboard to grapple with or help an interested party grapple with MS...."
All those brains in IBM, and screaming non-corporate users cannot get the time of day.
Intersting article.....but no mention of IBM making a serious mistake by trying to change to a proprietary bus on the PS/2 with their Micro Channel architecture (as compared with the open architecture of the original PC and then the AT).
Come to that, the article makes no mention of the PS/2 hardware at all. Perhaps a third part of the series could be on the relationship between OS/2 and PS/2???
The world desperately needed something to replace that shonky, old, nail-the-bus-lines-to-the-board-and-hope AT bus.
MCA was a far better solution than the horrific, fucked-up abortion of a compromised pig's ear that was EISA. The only thing IBM got wrong, with the benefit of hindsight, was not including a couple of ISA slots in MCA machines as became common (and still is) with PCI. I suspect that the chipset design of the period just wasn't up to it.
ISA was originally proprietary to IBM! Though the license fees to access the spec were low enough that it barely mattered, and they were eventually openly published, but it certainly wasn't that way from the beginning.
MCA might have succeeded if it had followed the original path, but instead they jacked fees up like crazy and adoption never got off the ground. It was 6 more years before PCI was developed, sadly, a time when paying through the nose for MCA was the only alternative to the slow buggy mess that was EISA.
MCA was killed off by marketing who wanted to sell boxes. I was at a talk by Chet Heath, the MCA architect where he talked about the "guest" card - an MCA card with a 386sx, a VGA chip and keyboard and mouse ports. Plug 3 into a PS/2 and you had a mini 4-person server. Cue massive panic in Sales. And so on.
... was trying to get OS/2 onto one.
I actually quite liked OS/2 1.1. It was a shame that
a) it was almost impossible to get it to install on computers, even IBM PS/2s.
b) it had a habit of lovingly resuming things on reboot so accurately that once it had crashed, it would crash in the same way on restart etc.
The deliberate confusion with having the extended edition allegedly only work on PS/2s was a big mistake.
I still have a copy of OS/2 2.1 in the shed. I never managed to get it to do any work though it installed OK, as I recall. BUT, where OS/2 did feature in my life was in the early 90s. We had a branch network consisting of mostly 9.6kbps analogue lines used to access green-screen Pr1me applications. The network was incredibly expensive - a million a year if I recall. So when email started to become useful in head office, it was decided that branch managers should have email access. (Don't you love the way technology is used to consolidate inflated views of status? Rarely is early technology used to help actual work gettnig done.) Anyway,. my boss paid a wodge of dosh to a consultancy who tried to get MS Mail to work via a Netware connection using ppp. They scraped 1000bps on the 9.6k lines. A mate at Microsoft suggested I try Lan Manager's remote access server - RAS. To cut a lobng story short, test showed a throughput 8 times better than the consultancy, and I implemented, installed and trained the users of the entire thing, complete with horrid hand-soldered serial cables for the RAS board, for less than the cost of the consultancy. OS/2 1.3 as a server offergin seemed pretty good to me, but it was seen as a Microsoft, not IBM, product. On the basis of that, when Netware became surreally priced and just too weird, we sailed wholesale into NT.
There are still a few ATMs with OS/2 around ( or there were 3 years ago ) 350MHz/32Mb/4Gb machines with OS/2 are snappy and responsive compared to the sluggish laggy 2GHz/2Gb/40Gb Windows XP Embedded machines which replaced them. Yes, the mechanical ATM parts are identical. Yes, the comms and backend transaction processing are identical.
If there are any readers in the mid-Atlantic states in the US: is it me, or does the Giant supermarket chain use OS/2 for their POS systems? Every time I go shopping I end up staring mesmerized at the remarkably OS/2-ish window chrome whilst the PFY scanning my groceries wonders what kind of nut job I am.
Hmmm, Escom when they launched in the UK had a deal with IBM. IBM staff trained the technicians at their pre-launch conference and every machine came installed with OS/2 Warp.
It was awful though, even in the tech class with IBM staff, it failed to correctly load on a number of machines and the IBM guys couldn't work out why. It continuously caused problems for Techies and Customers.
Having OS/2 Warp instead of Windows pre-installed made the sales guy's life harder, the techies life harder and the customer unhappy. Less than a year later they stopped installing and gave it as an option instead.
The OS might have been great for certain use cases and might have been far superior to Windows but it was a hard sell.
We replaced some generic PC boxes with Aptivas and they came preloaded with OS/2. Being game, we tried it and soon abandoned Windoze. We were developing PC-card (the ones from the PCMCIA) drivers, which were all DOS-based back then. Under Windoze, bugs took the whole box down. Under OS/2 Warp, you simply closed the DOS-box and opened up another one. Pure bliss compared to rebooting from an often damaged file system. (I also remember Borland's gui debugger -- the best I have ever seen.) We used the MKS Toolkit for OS/2 and had boxes pretty damned close to a real system.
From what I remember the two big failings in the design of OS/2 were graphics point-of-origin and supporting the brain-damaged 80286 processor.
Going for the GDDM point-of-origin (bottom left of screen) instead of the Windows poo (top-left) meant that Windows programs could not be easily ported to OS/2 (the simple transform need to check which of the VGA modes the graphics was in to convert poo -> which was unacceptably slow). The irony being that mainframe GDDM applications worked much better with IBM’s existing 3278 emulation cards & PCDOS than with OS/2.
Supporting the Intel 80286 was a mistake because while switching from real (8086) to protected (286) mode was simple, doing the other way required you to reset the processor and catch it as it restarted.. making task switching very slow. Windows never supported protected mode, cooperatively switching in real mode or extended mode in Win/386 (Win 2.1 for 386 processors). In hindsight IBM should simply have printed upgrade vouchers for the PC-AT customers.
OS/2 Extended edition was promised at the get-go because OS/2 communications manager with shared SNA connection was seen as the best way to displace 3270 terminals & database manager was no threat to big iron because it did not even include transactions at first.
No review of OS/2 is complete without looking at its role in bank cash-machines (where 3270 datastream to Stratus boxes was the norm).
Had it not been for IBM, we’d all have switched to Microsoft Xenix (the Unix OS MS originally pitched to IBM for the PC), and wouldn’t have had to wait so long of a decent UI on a *nix OS
I very nearly bought OS/2 Warp for my personal machine in 1997 or so, as I had already become sick to the back teeth of Windows 95's unreliability (this was win95a, on a P166 bought in 96) I was even seriously considering going back to DOS (DR-DOS, at any rate.) WordPerfect 6 for DOS was pretty good, had a mostly workable graphical mode if required, and most games still actually ran in DOS mode rather than Windows so it was worth considering.
In the end, the only thing that stopped me was hearing about Linux and installing that instead on a newly bought hard disk - I'm very glad that was the path I chose in the end; maybe it wasn't all plain sailing at first but what I learned in the process turned out to be very valuable indeed (got me a job, in fact) and is pretty much all still relevant now.
Interesting to read about OS/2 again from an insider point of view - I do wonder sometimes how anything useful ever gets produced by these corporations with their massive layers of "business" people!
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2023
