Why New York's Subway Still Uses OS/2 (tedium.co) 197
Every day 5.7 million people ride the subway in New York City -- and are subjected to both "the whims of the Metropolitan Transit Authority and the unheard-of reliability of a marginally successful operating system from the early 1990s."
martiniturbide shared this report from Tedium: OS/2 and MTA consultant Neil Waldhauer said in an email, "For a few years, you could bet your career on OS/2." To understand why, you need to understand the timing. Waldhauer continues, "The design is from a time before either Linux or Windows was around. OS/2 would have seemed like a secure choice for the future." So for a lack of options, the MTA went with its best one. And it's worked out for decades, as one of the key software components of a quite complex system...
Despite the failure of OS/2 in the consumer market, it was hilariously robust, leading to a long life in industrial and enterprise systems -- with one other famous example being ATMs. Waldhauer said, "Thinking about all the operating systems in use [in the MTA], I'd have to say that OS/2 is probably the most robust part of the system, except for the mainframe." It's still in use in the NYC subway system in 2019. IBM had long given up on it, even allowing another company to maintain the software in 2001. (These days, a firm named Arca Noae sells an officially supported version of OS/2, ArcaOS, though most of its users are in similar situations to the MTA.)
martiniturbide shared this report from Tedium: OS/2 and MTA consultant Neil Waldhauer said in an email, "For a few years, you could bet your career on OS/2." To understand why, you need to understand the timing. Waldhauer continues, "The design is from a time before either Linux or Windows was around. OS/2 would have seemed like a secure choice for the future." So for a lack of options, the MTA went with its best one. And it's worked out for decades, as one of the key software components of a quite complex system...
Despite the failure of OS/2 in the consumer market, it was hilariously robust, leading to a long life in industrial and enterprise systems -- with one other famous example being ATMs. Waldhauer said, "Thinking about all the operating systems in use [in the MTA], I'd have to say that OS/2 is probably the most robust part of the system, except for the mainframe." It's still in use in the NYC subway system in 2019. IBM had long given up on it, even allowing another company to maintain the software in 2001. (These days, a firm named Arca Noae sells an officially supported version of OS/2, ArcaOS, though most of its users are in similar situations to the MTA.)
Easy answer (Score:3)
Re:Easy answer (Score:5, Informative)
I installed OS/2 at some point I guess around 2000 or so just to take a look at it, and that is the extent of what I recall. Basically it was kind of interesting, but I saw no particular reason to change to it at the time.
By 2000, with Win2000, there is not much reason for any average user to switch to OS/2 anymore.
However, during around 1993-1995, OS/2 was pretty much to only OS that can run DOS/Win3.1 games/applications (until MS broke it with Win32s) AND allow you to smoothly multitask (such as full speed download from BBS or Internet). Quite a few BBS sysops ran on OS/2 so they can use their machine while running the BBS.
During 1995-1999, OS/2 beat Win95 in terms of stability, but most users didn't care and just reboot.
Re: Easy answer (Score:4, Interesting)
I used OS2 1.3 and 1.4 on 3 projects while working for CONRAIL. Believe it or not, the railroads are very technical shops.
The first project was for an interface to an Baerd oil spectrometer. It controlled the spectrometer and sent the results to the mainframes via LU6.2. The other 2 dealt with the loading, unloading, and management automotive railcars from the major automotive manufactures. At the time, sophisticated stuff.
Those apps ran 24x7 and didnâ(TM)t fail. Rock solid thanks to the OS.
Then, OS2 Warp come out. It was buggy as hell and hardware compatibility was sketchy. Then, they tried to make it run Windows 3.1 apps. It crashed all the time.
I stayed in the Windows ecosystem after CONRAIL was broken up into CSX and NS railroads. IBM had a chance. Lost to Microsoft by trying to be a jack of all things and master of none. Besides, MS had the game market.
Re: (Score:3, Informative)
Re: (Score:2)
During 1995-1999, OS/2 beat Win95 in terms of stability, but most users didn't care and just reboot.
Yep. I had it around that time for the desktop at home.
It was superior, in isolation, but it lacked software. Sure, it could run Windows stuff, but not all of it, not perfectly. At some point it just became easier to run actual Windows. The drop in stability was worth the tradeoff.
Re: (Score:3)
Similarly if you are looking for a reliable, dull but working Internet server with phenomenally wide range of hardware support, BSD family, especiall
Re: (Score:2)
NT 3.51 was a rock. In NT4 they merged the kernel and GDI memory spaces in pursuit of graphics performance. But today, Linux has better performance of all kinds than Windows any time there is parity between the drivers... Maybe they should have stuck with reliability instead of chasing performance.
Re: (Score:2)
I installed OS/2 at some point I guess around 2000 or so just to take a look at it, and that is the extent of what I recall. Basically it was kind of interesting, but I saw no particular reason to change to it at the time. I think I had a similar reaction to BeOS.
The biggest advantages for something like OS/2, BeOS or heck OpenBSD is they are cleaner operating systems that aren't trying to support everything, so are simpler and probably easier to secure. It takes less time to test things on simpler environments. The downside is they don't do as much. Still, if you have the hardware it will run on, an operating system will function the same way it always has.
It's just too bad that OS/2 always remained in the hands of IBM, even after the latter aborted plans to introduce OS/2 for PowerPC. Had that latter thing come to fruition, it could have helped IBM be more of a player in the desktop market when Intel and Microsoft were the only options. While they did change ownership ultimately, there is little reason for people to sink that money into OS/2 when they can get any Linux or BSD distro for a nominal amount, if not free.
Both the laptops I have today could ea
Re: (Score:3)
It's just too bad that OS/2 always remained in the hands of IBM, even after the latter aborted plans to introduce OS/2 for PowerPC.
OS/2 for PowerPC was pretty much what ended OS/2. IBM's development team wound up mostly being focussed on the PowerPC port, trying to get it up to par with the existing Intel version, instead of improving the Intel version much. For a time, IBM was telling customers and developers that PPC was the future of OS/2, stalling much int he way of investment in Intel based code and tooling.
And then it fell flat. Several years later, they had some test versions that were about as good as the Intel version of OS
Re: (Score:3)
OS/2 was more of a contender around 1990-1997ish (give or take a bit on the final year). But by 2000, Linux or one of the BSDs (in the end, it turned out to be Linux) was the clear winner and the obvious future for anyone who didn't think the purpose of their computer was to fuck them over. OS/2 had some basic non-shittiness that you couldn't find with Windows or MacOS 8-, but also had some clunkiness too, and it was still just as unmaintainable as those other proprietary OSes. Releasing it under a Free Software license was pretty much OS/2's only possible chance of staying alive, but they didn't do it (supposedly due to Microsoft owning the copyright on parts of it?), and so that was the end.
Just like Itanium killed the market for RISC, Windows 95 and 98 killed the market for OS/2, despite the former being delayed by 3 years. I was in college at the time, and the tech press - PC Magazine, PC World, BYTE, Infoworld, et al were always full of stories about 'Chicago' (the code name for Windows 95, for those of you Millenials) and its slippage. OS/2 was available, but at that time, it was pretty resource hungry, and Warp didn't seem to improve things much. So when Windows 95 came out, the opport
Re: (Score:2)
Re: (Score:2, Insightful)
I installed OS/2 Warp when it came out and played Doom on it. My Soundblaster card had a glitch and it required irq 7, not 5.
I'm not surprised they're still using software that Just Works (for them) but I have to wonder where they're sourcing the motherboards, chipsets, video cards, sound cards, etc. that are still compatible with it.
Re: Easy answer (Score:5, Informative)
Re: (Score:2)
hilariously robust? (Score:3)
Re: (Score:3)
Warp was horrendously unreliable when it arrived but prior versions were robust as all get out compared to the other main choice at the time (Win 3.x).
I loved it for years and was a strong supporter but I will admit Iâ(TM)m absolutely shocked itâ(TM)s still deployed in so many industrial use cases without being virtualized today.
Re: (Score:3)
If you need access to the hardware, virtualization isn't the best course. OS/2 is also the hardest OS to virtualize as it uses more features of the i386 then others, for example parts run in ring 2 to support DOS drivers.
And it does install and run on fairly modern hardware, though it needs a USB3 driver for those systems that don't have any real USB2 ports. Has other limits as well like 2TB disk and partition limits, which is to be expected for an OS designed in the early 90's.
Re: (Score:2)
Re: (Score:2, Informative)
There were two things which killed OS/2 as a mainstream desktop product or server OS.
First, companies were shying away from Netware. It was expensive and had strict license keys. Yes, you could run NetWare in an OS/2 DOS "coffin", but that was a cautious exception, as that required additional RAM and drive space (which were at a relative premium in those days.)
Second, people looked at OS/2 with the ability to run their Windows stuff, compared to just MS-DOS and Windows 3.x, and just decided to keep with W
Re: (Score:2)
I had the lovely experience of watching the OS Warp release with IBM execs (in South Africa at the time). The exec on stage actually greeted the Microsoft guys in the crowd. Then he proceeded to stick a PCMCIA card into his laptop and .... it crashed. The Microsoft guys and the rest of the crowd (rightly) laughed and found this amusing.
Re: (Score:2)
OS/2 2.whatever was delightfully reliable, unless you loaded the Windows environment into it. But there was little enough software for OS/2 that most people had to do that, so it was actually a serious problem for desktop use.
It makes as much sense to run it today as it ever did, however, if you aren't expecting it to be a modern OS, and it does the job you need it to do. I'm not sure how much sense it ever made, but I'm not spectacularly interested in arguing about it either. I imagine that there are few n
Re: (Score:2)
Warp was horrendously unreliable when it arrived but prior versions were robust as all get out compared to the other main choice at the time (Win 3.x).
Warp was fine upon release if you were installing it clean on OS/2-optimized hardware, and stuck with the most basic device drivers (like the standard SVGA driver), as you would on the types of industrial control systems the article is talking about.
It was only more problematic if you were upgrading an existing DOS/Windows system to WARP, and/or needed audio or better video drivers that it was somewhat problematic. But if you were using the system in a closet to run your print server or your voicemail syst
Re: (Score:2)
I've known research labs to use OS/2 for decades, however many were dedicated to specialized workloads. One ran in text mode, some were older versions.
Warp seemed stable to me at the time, depending on the PC, perhaps problems were due to drivers or that bolted on NetBIOS stuff...?
Re: (Score:2)
Warp (Warp Connect did) didn't come with NetBIOS, just enough of a stack to allow dial-up over SLIP (PPP support was released soon after).
What it was was particular about hardware. Your memory modules better be the same for example. I had a 486 with broken 16 bit DMA, it was really unstable until I switched to 8 bit DMA on the sound card.
Re: hilariously robust? (Score:4)
It had issues. The SIQ hang issues was one which was never really resolved satisfactorily but then it's not uncommon today for Mac OS to give me the spinning beach ball of doom because it also has a similar SIQ hang, 15+ years later.
Re: (Score:2)
LMGTFY
Synchronous Input Queue
Re: (Score:2)
I used to run OS2 for my multinode BBS, it was very stable until it suddenly crashed (which could take a fair amount of time), but when it did crash you'd lose basically everything (complete filesystem corruption, recovery from backups needed). This happened a few times, after which i just decided i had enough of it and switched back to DesqView.
I laughed at that comment too, but .... (Score:2)
Honestly, the things that were less than robust with OS/2 tended to be specific scenarios. You could certainly boot up OS/2 and leave it running for week after week with a basic installation, and find it kept on chugging. It wasn't like Windows '95 or 3.x where memory leaks would render the OS unusable after a while, without a fresh restart as a requirement. (As I recall, WIn 3.x used to have problems simply from the repeated process of opening new windows and closing them again.)
I had plenty of software
Taxpayers are dreading the follow-up announcement (Score:5, Interesting)
I've seen enough PR blitzes to know what's coming next: an announcement about the no-bid, nine-figure migration contract to a well-connected IT firm. Why DOES New York's subway still use OS/2? Because it still works, and they'd rather spend their money on pensions, salaries and occasionally rail equipment, that's why.
Re:Taxpayers are dreading the follow-up announceme (Score:5, Insightful)
Robust my ass (Score:2)
Re: (Score:2)
That was just the graphical interface. Text mode programs and such didn't have the issue and Fixpak 17 for Warp 3 mostly fixed it.
Re: (Score:2)
This was the OS that hanged if there was a poorly written program that didn't read its messages from the single input queue.
Which isn't an issue for industrial control applications, which probably only have a simple status UI, and is the only application running on the system, with little to no real user input.
And I'll note the entire OS didn't hang -- just the UI. Threads continued to run just fine, you simply couldn't interact with the UI when a SIQ hang occurred.
Not entirely ideal for end-user use (although it did get better in later releases), but for industrial control systems that run only one application and which don't
I had a copy back then (Score:2)
That thing was one seriously well implemented monster. Just serves to show that in the OS market quality is not something end-users can judge.
Re: (Score:3)
It was memory hungry at a time when memory cost a fortune. I ran it on a 386 with 4 MBs of ram, without the full graphic shell it ran pretty good, but really it needed 8-16 MBs when boxes still came with 2-4MBs.
Re: (Score:2)
Well, yes. But the competition was a bad joke in comparison. Of course, that was back then.
Re: (Score:2)
But the competition was good enough, ran on most hardware satisfactory if you didn't run too many programs and appeared to be free. And as most programs were targeting the competition...
"hilariously robust" (Score:5, Interesting)
"hilariously robust"
Yep, it sure was. This was back in 1996 when a friend's computer started to have random crashing problems with Windows. He couldn't afford to buy new RAM but OS/2 was somehow able to fence off the bad memory and continue to work reliably. Kinda funny that IBM effectively gave away OS/2 free on a magazine cover disk - applying a fixpack removed the modifications that IBM did to the coverdisc version.
Re: (Score:2)
My experience was that the OS/2 installer wouldn't even start with a bad memory even if Windows worked just fine.
Re: "hilariously robust" (Score:2)
Ok my friend's case, it was already installed. Given his machines hardware problems at the time and that he was skint, we didn't poke at what was working...
Re: (Score:3)
I (regrettably) yelled at someone who moved the mouse pointer out of the Windows window and onto the OS/2 desktop even though I specifically told them if they did that the system would crash.
It crashed. I yelled at them. I apologized for days.
IBM should have paid more attention to improving the OS and market development. Sure, it was designed to run its "userspace" in subsystems, like Windows NT Windows-on-Windows did much later, but supporting MS Windows it sucked the life out of the project.
Memory protection on commodity hardware (Score:5, Interesting)
Good development tools, for its time, and proper memory protection on i386 made it a good choice for industrial controls in that era. The main drawback to OS/2 today is the lack of support from the original vendor and that the development tools haven't really kept pace. You do see NT kernel and Linux in the same sorts of roles more frequently these days than OS/2, and it may be as simple as it is easier to hire engineers with experience in Linux or Microsoft Windows than with OS/2.
I'd label OS/2 as not for new designs. But just because software is old doesn't mean it magically loses the ability to meet its original requirements. If it ran a system fine 20 years ago, and the requirements haven't fundamentally changed, then it can continue to fill that role.
Re: (Score:2)
The biggest problem is lack of kernel source, and its a hard kernel to binary patch. But you can buy a decent computer today and as long as it still emulates the BIOS and has a real USB2 port (or you add a card), it'll install and run. Even give you a ram disk to use the memory above 4GBs. HDs and partitions are limited to 2TB as well due to CHS addressing. Limited to 64 cores (I think only tested with 32) as well.
Re:Memory protection on commodity hardware (Score:5, Interesting)
OS/2 found a lot of its niche during the 90s for those things that needed "more than a desktop" but "less than an AS/400" - because in the 90's, it was still pretty true that nobody ever got fired buying IBM.
TFA cites ATMs as being one of those places where OS/2 was used well after Windows ruled the desktop, but the reason wasn't necessarily because the banks wanted to give it up. After all, a computer with a a cash dispenser is probably not a good thing to have running Windows in the era of Blaster and Sasser. However, ATMs started running Windows for a very different reason: the Americans with Disabilities Act. Before the ADA, ATMs only had to comply with whatever the insurance companies were requiring, and OS/2 was on the whitelist. The ADA required that ATMs be accessible to blind people. A fantastic thing to address, but "good idea + government bureaucracy = worst possible solution". I'm sure it's trivial to write a screen reader program for OS/2, there wasn't one for the OS that met the legal criteria, meaning that banks had to start rolling out ATMs that could run the required applications for blind people, which meant that they had to start running Windows.
Re: (Score:2)
If they'd bought an AS/400, they could have upgraded and/or replaced outdated hardware at any time, and still run their applications to this day.
AS/400s rock. They just keep working. They ain't fast, but they don't stop.
Re: (Score:3)
If someone bought an AS/400 these days, they would have an upgrade path today to an IBM iSeries. Yes, the old CISC CPUs were poky, but you really didn't care about that when running a business, and now that AS/400s run on POWER9, there are no performance issues these days.
Of course, the downside is price. IBM stuff doesn't come cheap.
Re: (Score:3)
IBM stuff doesn't come cheap.
When you factor in the cost of man-hours maintaining and porting software to new generations of hardware, you will find that the upward-compatible family of systems designed by IBM Rochester (System/36, System38, AS/400, System i) is a huge bargain.
Check out Frank G. Soltis' book Fortress Rochester: The Inside Story of the IBM ISeries for details.
Re: (Score:2)
The current weakness of the OS/2 of today is development tools. If one is making an embedded app, they will be using Linux, an embedded Windows edition, QNX, seL4, INTEGRITY, or another OS that is constantly maintained and updated.
It would be nice to see OS/2 open sourced somehow. I'm sure there is a lot of entangled IP in the code, but with a clean room rewrite, the code likely could be made into something that is more usable. From there, add tools for various embedded applications (kiosks, digital sign
Re: (Score:2)
There is a performance penalty to traversing through the protection rings and to using TSS as intended. And segmented protected mode is not comfortable to program, especially in environments that want to use libraries that have full access to the same address space. Using these x86 features is not the fastest way to use an x86, but it is the most mainframe-like way to use it.
Hardware is the issue, not the software. (Score:2)
So long as your I/O ports are hermetically sealed then it's fine to run an old stable OS like OS/2. However, the real problem here is if your stable software will only run on old hardware that was not designed to run for decades on end. When your hardware fails then you are up a creek without a paddle... but they will sell you one for a very pretty penny.
To avoid being held captive, they should to buy an embedded system that is designed to run for eternity and then either port the software or run OS/2 in
Re: (Score:3)
You just buy a new box. Neil who is referenced in the summary can set you up but basically you need good hardware, BIOS emulation and some USB2 ports. And need the latest AOS (ArcaOS) which is basically OS/2 4.52 with updates.
Re: (Score:2)
Intel still sells the 286 and 386s. Industrial computers haven't changed in pricing or features for that matter. Sure you can get brand new Industrial Computers with Corei7 but the heat management of a TDP 180W chip vs a 10W 386 chip is immense in these situations.
Follow the money (Score:2)
To pay for decades of other unrelated city services.
It could have been the better Windows (Score:2)
... and a valid alternative to Linux on server platforms, too.
Because of imaging needs, I gave up on OS/2 on the desktop back in the early noughties, but I continued to run an OS/2 machine as a production internet web/mail/file/application server until two years ago. It was mainly errors and stability problems in OS/2 ports of necessary standard software (Apache/PHP/MySQL) that forced me to go to Linux, as compatibility with PHP webapps had become a game of luck, even making frequent reboots necessary due t
Re: (Score:2)
The OS/2 UI, like the older MacOS UI would never freeze. You could copy a floppy from A: to B: (or later on burn a CD at 4x) and the UI would still be responsive, the hard drive (if you had SCSI) would also be responsive even when you (later on) had a CD copy going on, Windows would be completely unresponsive during that operation or fail burning the CD (remember that time when you had Nero reported a "buffer underrun" and you had to chuck your $1 media carrier).
That's why I used OS/2 until the late 90's an
Sigh. (Score:3)
Because, obviously, you wouldn't buy a billion-dollar MTA system with hardware integration that didn't include complete specifications of the hardware interface and necessary protocols/signals, so that you could easily start to replace small parts of it bit by bit, testing it against the original implementation until you were happy that it performed the same actions, and then start to replace parts of the system with something vaguely modern and compatible and continue to be able to do so into perpetuity, would you?
I mean, nobody would be that stupid, right? To be tied into the exact piece of software that was given on deployment for systems that - in other countries - traditionally operate for decades, even centuries at a time?
Hell, even places like the London Underground revamp their computer control suites and signalling every decade or so and they're always pleading poverty. And that, quite literally, been running for over a century (though, obviously, not computer-controlled all that time).
I get the "if it ain't broke" mentality, but the thing is that at some point I guarantee that it *was* broke... and they just kept buying the same parts over and over and over again at great expense. And even on the tiniest scale, you could have had a plan in place after the first few of those to code up some other replacement.
Hell, you could probably run the entire thing on embedded devices nowadays. Get the damn things custom-made, so you're holding the blueprints and have no real "software" to deal with beyond those bits you want to.
I'm not suggesting IoT the thing, but you could at least have started to say "Let's contract out producing a modern replacement for this control machine and its software."
Re: (Score:2)
Wow, you've got all the answers. NYC should hire you.
Re: (Score:2)
Thanks for letting us know you are clueless about industrial system. Would have saved everyone a lot of time if you just lead with that.
""Let's contract out producing a modern replacement for this control machine and its software.""
It's adorable you think that's all that is involved. Like the running system won't rely on anything else but that.
Simply, adorable.
"but the thing is that at some point I guarantee that it *was* broke"
oh, do you now? lol.
Mainframe connectivity (Score:2)
I loved OS/2 (Score:3, Interesting)
When I switched from OS/2 to Windows 95, at first I was delighted by Windows 95's new user interface, with the Start menu and taskbar, which were not common at that time. Long filenames had been used in OS/2 for years, as was preemptive multitasking. But OS/2 had amazing settings for its applications. A graphical representation of a notebook grouped all sorts of settings on different "pages" of the spiral bound notebook. Underneath, OS/2's extended attributes allowed all sorts of data to be stored for each file. And OS/2's system object model was much, much better than Windows' common object model. Move a file on OS/2 that had a link to it; the link automatically updated to the new location. Do the same thing in Windows 95, and it would show a moving flashlight icon as it spent a very long time trying to search the hard drive for where the target file had been moved.
After various versions of Windows kept crashing, and after my Windows XP machine would not boot up because I (gasp!) uninstalled too many programs at once without rebooting in between, and my Windows XP registry subsequently got corrupted, I finally came to my senses and switched to Linux around 2002. I have run Linux as my primary operating system ever since and never looked back.
I still miss OS/2, however.
Re: (Score:2)
No Netware backend? (Score:2)
Re: (Score:2)
Novell did run on OS/2. I had an IPX/SPX network operational on OS/2
IBM chief: Microsoft killed OS/2 (Score:4, Informative)
The deposition and testimony provided by Garry Norris - IBM's chief negotiator with Microsoft before and after the introduction of Windows 95 - has provided a cornucopia of fascinating evidence in the Microsoft trial. Much of it was previously unknown or unconfirmed.
His evidence showed how Microsoft effectively controlled IBM's PC hardware and software businesses by making the price of Windows considerably higher than for other comparable PC makers.
Mr Norris described in detail to Philip Malone, counsel for the Department of Justice, five cases where Microsoft had succeeded in modifying, or had attempted to influence, IBM's choice of software to load on its PCs. These were with OS/2, Lotus Notes, Lotus Smartsuite, World Books, Netscape's browser, and the version of Windows.
IBM's competitive operating system OS/2 was a significant challenger in 1994 and 1995 to Windows 3.x and then Windows 95.
As a consequence of its alleged Windows monopoly and IBM's need to include Windows with some PCs because its customers wanted it, the licence price has grown and grown:
In 1995 IBM paid Microsoft $40m for Windows
In 1996, after the launch of Windows 95, it was $220m
By 1997, after Microsoft insisted that IBM would have to pay higher prices if it wished to license Windows NT 4, IBM paid Microsoft $330m for Windows
In 1998 this increased to $440m
Mr Norris said that PC manufacturers who wanted to license OS/2 were threatened by Microsoft. Compaq was one of these, and IBM was informed of Microsoft's intimidation by Compaq vice-president Mike Clark. As a consequence, Compaq did not license OS/2.
It was revealed that Bill Gates was "surprised" that IBM was not prepared to "jointly and exclusively promote Microsoft products [and] reduce shipments of OS/2".
Re: (Score:2)
I worked for a vendor who was threatened by MS sales people that they would pull licensing if they also sold OS/2
OS/2 was great (Score:2)
Fond Memories (Score:2)
obsolescence is often a myth (Score:2)
We obsess over the latest and greatest in this industry, and certainly in some fields that is very important, but, not unlike farming, sometimes the newest tool is best and sometimes there's little advantage over that old tractor on the barn.
That said, while I was interested to hear the MTA is still using OS/2, the linked article itself is pretty tedious (as advertised), spinning a sentence's worth of information over many paragraphs, something that seems obsolete in the age of Twitter.
Mature (Score:2)
Bad software ages, good software matures.
No reason to through out working software for the sole reason of "IT's old".
Re:umm before windows? (Score:5, Informative)
Not really. I suppose it depends on which version you are talking about. OS/2 1.0 was text only & predated WinNT by quite a bit.
Re: (Score:2)
Re:umm before windows? (Score:4, Informative)
No, OS/2 v1.x was mostly written by MS under contract with IBM. Then they decided that IBM would do version 2 and MS would do version 3. Version 3 first appeared as OS/2 NT, but by 3.1 was Windows NT.
It ran OS/2 1.x programs up till W2K and 1.x graphical ones if you added the Presentation Manager kit.
Today the latest version of OS/2 is 2.45 internally.
Re: (Score:2)
Re: (Score:2)
I have a Byte magazine around here somewhere that has a brief news clip about Microsoft first booting up OS/2 NT v3 (or OS/2 v3 NT). Only text mode and running on some MIPS processor IIRC, but it was first booted as OS/2. Of course by the time it was done, it was Win NT 3.1.
Re: (Score:2)
This factually incorrect for any reasonable interpretation of "mostly written".
OS/2 v3 NT renamed Windows NT (Score:5, Informative)
Uhh, OS/2 was just IBM's version of WinNT
OS/2 1.x was a 16-bit contemporary of 16-bit MS-DOS. Except it required a 286 and ran in protected mode.
OS/2 1.x + Presentation Manager (PM) was a contemporary of MS-DOS + 16-bit Windows 3.1. At the time Microsoft and IBM were collaborating on OS/2 1.x and Microsoft said that Windows was just a temporary thing for DOS users who couldn't or did not want to upgrade, and that everyone who could should move from DOS to OS/2, that DOS was a dead end and OS/2 the future. Again, Microsoft was saying this. The programmer APIs were extremely similar. Porting from DOS+Win to OS/2+PM was fairly simple.
OS/2 2.x was a 32-bit 386 based rewrite that offered a protected multitasking environment before Windows NT shipped. It was written by IBM.
OS/2 3.x, aka OS/2 NT, was written in parallel to OS/2 2.x by Microsoft. It was the CPU portable version that ran on non-x86 systems. During its development IBM and Microsoft stopped collaborating and Microsoft rename OS/2 NT to Windows NT.
Re: (Score:2)
OS/2 3.x, aka OS/2 NT, ... the CPU portable version that ran on non-x86 systems.
and x86 too.
Re: (Score:3)
Re: (Score:2, Interesting)
Dave Cutler developed NT for Digital Equipment Corporation, and DEC did not care. Cutler left Digital for Microsoft in October 1988, taking his work and most of his collaborators with him. Microsoft did not develop NT, DEC did and it was stolen when Cutler left and took it with him.
Re: (Score:2)
Re: (Score:2)
That explains the bloat, then.
I had Win NT 4.0 running on a re-purposed DEC Alpha for *years*. The box's first life was a file+print server for PC clients using DEC client access licences (can't remember the actual name, but they were paper-only, and EXPENSIVE), then it was turned into a Windows Exchange server. It was three or four years into its new life before a disk controller went titsup. The Accountant wanted to know the cost of repairs - I was able to convince him it was time to move on. We replaced
Re: (Score:2)
Re: (Score:3)
It was still OS/2 after Cutler came aboard. Before and after Cutler OS/2 3.0 NT was never based on OS.2 2.0. It was always a complete rewrite with CPU portability and a new architecture, among other things. Hence the NT, "New Technology".
Correct. In particular, VMS + 1 = WNT.
Compare the architectures and APIs.
Re: (Score:3, Interesting)
I have a friend who was a young IBM EE with decent software development skills. He was working there at the time of OS/2 and he learned that the managers in charge of OS/2 - with zero experience in software - were told by Microsoft that the best language to develop OS/2 was assembly language . The managers bought the bait hook line and sinker, and a disturbance in the force was felt as if a million
Re:OS/2 v3 NT renamed Windows NT (Score:4, Insightful)
... the managers in charge of OS/2 - with zero experience in software - were told by Microsoft that the best language to develop OS/2 was assembly language. The managers bought the bait hook line and sinker, and a disturbance in the force was felt as if a million IBM software developers cried out in agony all at once. That move brought OS/2 development to a crawl, and Microsoft quickly grabbed the market with Windows 3.1
Given the Win3.1 reference I assume you are speaking of OS/2 1.x. Microsoft and the IBM managers were correct. Assembly Language was the better choice for the operating system. MS-DOS was also written in assembly. Speaking as someone who wrote assembly, C and C++ in those days, and someone who liked assembly despite the indoctrination and hatred and fear BS and MS CS programs attempted to instill.
Its unlikely that much of OS/2 2.0 was written in assembly language. Kernel and driver stuff mostly, and perhaps a spot or two of performance critical code. Recall that OS/2 2.0 was by design 386 based. It was only OS/2 3.0 that was to be CPU portable and would thereby have much more reluctance to use assembly.
IBM's grand plan was to monopolize the market with OS/2 shipped standard on every PS/2 but when the market went flat they not only killed the PS/2 line they also killed future development of OS/2.
I'm not sure I buy that PS/2 killed OS/2, certainly it hurt but both IBM and Microsoft were promoting OS/2 as the successor to DOS for all PCs, not just PS/2. What probably hurt more was hardware vendors not bothering to do OS/2 drivers, just DOS drivers. This helped create a hardware compatibility issue. Its the same sort of problem that Linux had for so many years and likewise found as an impediment to desktop success.
Re: (Score:2)
Re: (Score:2)
Why did they go w/ Mach for the kernel, given that the theory of microkernels was still unproven, and OS/2 was not Unix? Couldn't they have simply recompiled OS/2 Warp to the PowerPC, do a deep dive at CPU assembly level into the parts of the OS that were written in x86 assembly, and go on? What were the other challenges in bringing that to market?
Re: (Score:3)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Stone age tech?
http://3.bp.blogspot.com/_qab8... [blogspot.com]
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Linux and Windows were well established when OS/2 came out
Not sure that's right. Linux first version was early 90's. OS/2 was available in the late 80's. Taking into account that Linux then would also be pretty esoteric. Redhat wasn't formed until later in 1993, so I doubt Linux was a sane option at that point. Windows NT was also 1993, so at that stage OS/2 would have been stacking up against 16-Bit windows on top of DOS, rather than windows as most people would recognise today, so stretching it a bit, but in reality neither Windows or Linux would have been that
Re: (Score:2)
Re: (Score:2)
I still think SMIT for Linux would be a good thing for someone starting out. AIX has some very cool features, and PowerVM is definitely a nice choice, although not cheap. The features I like most about AIX are trustchk (allowing you to have the OS sign not just executables, but scripts and libraries), getting rid of root altogether and using roles (similar to Solaris), the ability to back up the whole OS with mksysb or sysback and restore it, and excellent binary compatibility. If I want to, I can run an
Re: (Score:2)
I still think SMIT for Linux would be a good thing
I think SMIT would be an excellent addition to Linux. The best thing about SMIT in my opinion, is the way that it is essentially self-documenting. At any time after you have it create a really nasty command line, you have the actual command that was executed right there in the log. I used it quite a bit many years ago when doing work with LVMs and similar things where you ended up with big hairy commands that would do exactly what you wanted.
Re: (Score:2)
Yeah, I had to rewrite Linux SCSI stuff to get it to work properly, so 'pretty new' is quite the euphemism.
Re: (Score:2)
Warp came with Win3.1, but it was always struggling.
That depended on which edition you got. If you got the "Blue Box" edition, it did indeed come with an IBM compiled version of Windows 3.1. However, if you got the Red Box Edition, it didn't come with Windows at all, and would use the Windows 3.1 installation from Microsoft already present on your system.
(There were plans at once point in time for a Green Box edition. I used to have a marketing slide from IBM on it, but can't remember off the top of my head what it did differently. It may have been an ea