Slashdot Mirror
25 Years of IBM's OS/2
harrymcc writes "On April 2nd, 1987 — 25 years ago today — IBM announced OS/2. It was supposed to be the next-generation operating system that would replace DOS. It never did. But for a famous failure, it's doing okay — it still runs the computers that manage the New York Subway's Metrocard fare cards, for instance. Over at TIME.com, I've taken a look at its occasional triumphs, frequent tribulations and enduring legacy."
In 1995, OS2 desktop was as popular as Macintosh. Now the field is pretty much 85% Windows with 10% Mac and under 2% Linux.
When the foot seeks the place of the head, the line is crossed. Know your place. Keep your place. Be a shoe.
I spent may hours working in the ICLUI interface building apps for OS/2. For the most part it was good at memory management, tools were mature and the interface was object oriented. I was always frustrated about the MS & IBM split on the interface and I think MS took the wrong route in getting to Windows. Had the alliance stuck around who knows what would have happened to this OS.
Audi, vide, tace, si tu vis vivere
I ran OS/2 extensively from '93 to '03. OS/2 was way ahead of it's time in many ways - maybe too much so. It was a great solid system and the GUI was much better than most of what we have today. it's a shame that IBM couldn't market it properly but they were working against the massive marketing force that MS had back then. That, and the fact that OS really ran best with at least 16mb or RAM back in a time when 8mb was considered excessive. Once Win95 came out OS/2 was pretty much on a fast path to it's death. That clearly demonstrated that the PC industry was more about marketing and deals than producing a better product because windows 95 was absolute trash in comparison.
Alex, I'll take keybindings not used by Emacs for $400....
"More than 250 companies declared their intention to deliver OS/2 apps, including biggies such as Lotus, WordPerfect, Borland and Novell."
OK, that made me smile.
Had excellent scripting, good multitasking, was very stable at the time compared to just about anything that you could run on PC hardware. I also remember it as being very fast, unless you ran Windows applications on it.
IBM was just not flexible enough to win. The exact same thing is happening to Microsoft right now with the only difference being that while IBMs desktop efforts died with very solid products at hand, Microsoft falls on their nose with crapware. Dont get me started on the duct taped Windows Phone 7 GUI with dripping glue onto Windows 7 that is called Windows 8. Every single engineer involved in that crap should be ashamed to the bones.
HTTP/1.1 400
OS/2 is still being updated/supported just not as OS/2. eComStation Is currently available and works with most current generation hardware.
"A person is smart. People are dumb, panicky dangerous animals and you know it." - K
CONFIG.SYS
Well, there's a longer story. Anybody interested should look into the blind luck and frustration that led to MS building Windows as "PM lite" and chancing into Dave Cutler's expulsion from DEC. The book "Big Blues" is a decent start.
When IBM pivoted hard toward PS/2 and 16-bit computing, Gates took one of the 3 or 4 intuitive gambles that defined both his success and that of Microsoft.
There's ONE simple use case, that illustrates the technical failing of OS/2, vs Windows NT - particularly in face of the claim IBM made for a "Better Windows than Windows". > > >. OS/2 didn't perform a special trap for that key sequence. Nor could it - without the 32-bit native, 'Virtual 8086" mode of the 386 processor. This simple illustration exposes the huge architectural gulf that OS/2 was unprepared to cross as 16-bit. Bill's certainty that 32-bit architecture was demanded by multi-task/multi-user computing in 1989 paid off. Inheriting the VMS brain-trust allowed him to execute, while leveraging the design and code contributions his team had made to the OS/2 project.
Besides that? CONFIG.SYS. Really! A whole /etc directory reduced to the parsability of one file! In this context, the follies of the Windows registry appear to be, comparatively enlightened.
"Flyin' in just a sweet place,
Never been known to fail..."
Up until recently, the teller terminals at my bank ran OS/2, but it was basically just a platform to run terminal sessions to the mainframe. Then they switched over to a browser-based front end to a UNIX back end.
The company I work for still supports old, legacy, OS/2 systems used for telephone menuing systems. It's funny that when there is a problem, many of the employees we support have no idea where the machine is located. It was literally stuck in an office somewhere and has been running completely unattended for years. It never gets updates. It never has to reboot. It just runs... and runs... and runs.
The problem we have now is finding hardware old enough to support it. We have to use 80GB drives for replacement and set up a 2GB partition for the OS and software. The rest just sits there idle. AT motherboards, ISA graphics and PS/2 keyboards and mice are getting harder and harder to find.
There is no "I disagree" mod for a reason. Flamebait, Troll, and Overrated are not substitutes.
Microsoft used to update win32s every week it seemed then IBM would fix OS/2 to run them. Finally with Win32s v1.30 Microsoft hardcoded some DLLs to load in high memory and as OS/2 only supported 512 MBs per process, no more Win32s support without a lot of work.
https://en.wikipedia.org/wiki/Inverted_totalitarianism
I was in college at the time and did a co-op with IBM that year.
My group used to call you CO-OP guys "NOP"'s - no operation - as in assembly 'NOP'.
You were easy to pick out - shirt and tie for the first week on your NOP job.
Yeah, yeah, yeah, we gave you punks a hard time, but it was out of love, man. You were sharp and ambitious and would end up as our boss. We had to take our shots while we could.
I still have that T-Shirt.
Me too.
I was in my local NAPA auto parts store and this old guy (even older than me) saw that shirt and said, "That's a really old T-Shirt."
Long story short, he was one of those guys that took an early retirement.
When I was at Boca, I watched all those "out of date loser" mainframers come down from NY to do shit jobs. I smugly thought, "That's what you get for not staying current!"
How arrogant I am. And I'm ashamed for it.
I escaped to a so-so business back office programming job while others were poached by Microsoft - the smart ones which wasn't me (Peter, peter rice eater - you rock man! I hope you're a MS Millionaire because you deserve it!).
The ironic thing is that the Hartford Insurers (who still train, btw) need some mainframers.
I met the most obscenely talented and genius people at IBM.
Looking back, it was the most humbling experience ever - and I was too arrogant to take that lesson in at that time. Then again, we have to be arrogant to get jobs in this fucked up industry, don't we? Saying, "I don't know." is the kiss of death.
Unfortunately, I never owned a PC during the time that OS/2 was around, and so never got to experience what it was. But most of the people who ever used it liked it. Just hearing about some of the concepts - dragging a file to a printer icon in order to print - blew me away. An OS that would have been the offspring of OS/2 and NEXTSTEP would have been just purrrfekt!
In college, I learnt about microprocessor design on a PPC 601 - the first PPC to come out after IBM did a derivative design of it along w/ Motorola (now Freescale). Knowing that OS/2 was going to have an uphill battle outside IBM (heck, even Amber didn't offer the OS), I was rooting for OS/2-PPC, which was known as Workplace OS. Unfortunately, as it turned out, Mach 3 turned out to be a horrible choice for a kernel (and Hurd pretty much made the same mistake in going w/ it) and finally, IBM canned it. That was the real death knell of OS/2, and w/ it died any real hopes of the PPC getting popular outside Apple (as far as computers go - I'm not thinking about consoles or other boxes)
Incidentally, today, there is a project called OSFree, which is similar in concept to Workplace OS, except that it uses the more recent L4 micro-kernel as its underpinning. The concept here is good - on top of the micro-kernel, they plan to use different 'personalities', such as Presentation Manager, Win32, DOS and even Linux (there already exists an L4Linux, so they may not do much more on that one), as well as a Neutral personality, which would provide the services that the other personalities require. The advantage here is that the portability of the L4 has already been demonstrated, since after an initial design w/ some assembly code, it was found that replacing assembly code w/ C didn't have any performance impact.
I know that at this point in the game, computers based on anything other than x64 or ARM are pretty much non-starters, but it would be fantastic if such a project actually came to fruition. That would be a good step towards portable computing, while giving just about any architecture the ability to have an environment like OS/2. Hopefully, all the major FOSS software will be ported there, and that platform would then have a chance of being viable. I think that b/w OSFree and ReactOS, there should be enough opportunity for OSs that decide to take advantage of the end of support for XP. Maybe a laptop based on a MIPS or PPC can have a go at it
I remember it well, I was tasked with a number of OS/2 projects. Recoding MS's writelog function, making it asynchronous (non-blocking), creating the (AT) VGA driver, creating (AT) ST506 driver, and the biggest challenge ever, I was tasked with creating the final quality control steps/code/testing methodology.
I knew the final QC phase would be huge, an almost impossible challenge, since the Microsoft's core staff was mostly Recent College Grads who would take many of inappropriate shortcuts. Thus it would take something extraordinary to beat their code into something useful.
If I had failed, I suspect the micro computer industry would have been stuck in a dark age for at least a decade, maybe more.
The biggest hurtle was there would be no way to fully test all combinations of system functions, our SUN would burn out first(billions of years). Instead of attempting the impossible, I did the next best thing.
I created a series of revolutionary stress tests for that project. The component programs were a series if self checking programs which used out of phase pseudo random number generators. The resulting (re-creatable) data patterns were used for both the function parameters and content, and the longer they executed, the greater the testing coverage.
Long story short.. The first release of OS/2 (86) never saw the light of day.. It couldn't even pass the individual component stress tests, let alone dozens of them in combination, all controlled by my screen manager. Sloppy coding techniques and shortcuts had forced MS coders to go back to drawing board and start over from scratch.
Net result, those stress tests uncovered many flaws, including hardware problems, and major software issues, some of which were carry overs from PC/MS/DOS. They were discovered and fixed, some of them were folded back into next release PC/MS/DOS, 4.0. Thus making DOS based PC's useful for large databases for the very first time.
In the end, the code, the methodology I created, was so far ahead of everything else they quickly took over all other forms of OS testing at both IBM and MS. And it lives on to this day, Microsoft has ten's of thousands people creating/running modern permutations of those 24hr stress tests I pioneered for the birth of OS2, using it to find and fix bugs in all versions of windows.
Despite all that if you knew what you were doing it was far more superior to anything Microsoft had at the time. I'm sure Microsoft engaged in all kinds of sharp practice but it really needn't have bothered. IBM was its own worst enemy. By the time NT4.0 / W2K were appearing there was no reason at all to use OS/2.
IBM engineers had the full Win32 running on OS/2 but once Microsoft found out they modified Windows 95/aka Chicago to break that capability. OS/2 processes could only access 1GB of address space while Chicago processes got 4GB of address space. So to break the OS/2 ability to run 32bit Windows Microsoft modified their resource compilier to load the applications resources(menus, icons, etc ) up at the top of the address space instead of down low with the rest of the application. Viola, OS/2 was unable to load the full Win32 application.
There were stories of IBM even solving that problem but deciding that if Microsoft was willing to convolute their OS design to prevent OS/2 from running it once, they'd just keep doing it and so IBM ended the cat/mouse game at Win32S capabilities along with OS/2's already advanced design.
LoB
"Anyone who stands out in the middle of a road looks like roadkill to me." --Linus