I developed for OS/2 in the 90's. OS/2 was 5 years ahead of Microsoft Windows, and offered full 32-bit processing long before Windows did. However, OS/2's strength was also its weakness.
OS/2 was developed from scratch, and was not encumbered by a legacy of DOS compatibility, 16-bit applications, and drivers like Windows was. This enabled IBM to innovate radically and more quickly. The application I wrote for my client was a multi-threaded program, written paradoxically with Microsoft's Programmer Workbench, the character-based precursor to Visual Studio.
However, this freedom from legacy code also meant there were few applications available for OS/2, and the only drivers were for IBM printers. OS/2 eventually disappeared in spite of its technological superiority at the time.
IBM's development environment was called C-Set, and, despite being a 32-bit compiler, was inferior to Programmer's Workbench. For instance, after a compilation, if you double-clicked on an error message, it would open up the source code in a separate editor window, in a proportional typeface. You could not trigger a compilation from this window. You had to make your change, save & close, then go back to the compiler window. Awkward.
There was no Resource Workshop for designing dialog boxes visually. You had to write all the statements in a text editor, and wouldn't see the appearance until compiling and running the program. This slowed down development.
While debugging, I recall trying to step into the GUI part of the application code. This locked up the computer so bad we had to re-install OS/2, a tedious process requiring 22 diskettes.
The IBM documentation was so inscrutable that even for ver. 2 of OS/2, we had to refer to Microsoft's ver. 1 manuals. IBM's support gave an interesting insight into OS/2's downfall. My client had an expensive OS/2 support contract with IBM. The client contacts were two guys in the PC support section. When I required technical support to answer some obscure and specialized questions about OS/2, it didn't make sense to get these guys involved, even though they were supposed to be the only official go-between with IBM. So I called IBM with the client's authorization, and impersonated one of the support techs. After a few calls I slipped and used my name. The IBM guy was livid and tore a strip of me, saying that as an independent contractor, I had no business blah blah. Here I was just trying to get a customer up & running.
I worked on an IT helpdesk at an insurance company for the summer in 1996. I had to support IBM PS/2 PCs that would tri-boot Windows 3.1, OS/2 Warp, and Dos. No one operating system had all of what they needed.
Funny side story:
They had a token ring network, and in the network room, there weren't enough physical network ports for all of the employees. Every morning someone would call the helpdesk to say their computer was couldn't access the network, I'd look at the N network cables, find one that without lights to indicate that cable was in use, unplug the inactive one, and plug in the cable of the person w/o access. It turned out on most days enough people called in sick or were on vacation that we could get everyone who needed a connection wired up :)
Anything that decided to hit the hardware directly from DOS, or did something that OS/2 didn't support in windows just crashed or had undefined behaviour.
Even windows in protected mode couldn't do all DOS things any more. Even if you dropped out of protected mode, and fiddled around too much you'd end up breaking windows when you switched back to it.
Back then you'd have all manner of weird DRM schemes which used dongles on the parallel port, not to mention manually set IRQs, I/O Ports and blocks of high mem (between 640K and 1M) where your device drivers were baked into the DOS application, NOT the operating system. There were some driver.sys type drivers for random hardware, but mostly it was just cheaper to write the driver inside the app itself.
Programming under DOS was more akin to writing to an Arduino rather than linux/windows back then.
Good times. I also ran OS/2 from v1.1 through OS/2 Warp.
Not quite true about the 32-bit no DOS. OS/2 1.x was a 16-bit system (although protected mode with segment selector 16MB memory) and one real-mode fullscreen DOS box. You're speaking of OS/2 2.0 which was a combined IBM and MS effort with OS/2 3.0 ultimately becoming WinNT. OS/2 2.0 even ended up having Win16 compatibility via DOS VM.
My favourite version of OS/2 was 1.3 with the SVGA driver written by an intern at IBM.
Edit: about the lack of GUI editor. My first paid OS/2 project was to write one after 2 weeks at a MS OS/2 campus.
It's all about the drivers. No OS will get seats without drivers. People always say that no OS will get seats without applications, which is also true. But because people see that problem, a big company trying to launch an OS, or a migration, will go out of their way to get the important applications.
The lack of drivers is more under the radar. So then people are shocked when their shiny new OS doesn't get uptake because too little hardware works with it.
Speaking of drivers.... didn't OS/2 drivers run at ring 1? The X86 architecture (and only a few others) support more than 2 rings of security. Most OS's don't bother using more than ring 0 for kernel and ring 3 for user space because then portability will be hamstrung. But OS/2 did not have the BSOD problem of Windows. By far the most common BSOD in the Win95 days was a 0x0E exception -- page fault in the kernel, the fault of a buggy driver that forgot to wire down a page before doing a slow I/O operation. A page fault in ring 1 is valid, and the kernel can handle it.
Seems like a challenging project. I remember OS/2 fondly, and I remember (and have heard tales of) its stability. Still, I have read stories that claim the source code was a hot mess of hacks and liberally sprinkled assembly code.
Apart from “hacks and glory” (which for the record is a fine justification), I don’t imagine that this project would have a lot of utility. While some old ATMs and some other legacy applications run on OS/2, I’d guess that attaining the degree of compatibility needed to satisfy those applications would be an extremely tall mountain to climb.
I think there are a lot of us who remember computing of the 1990s and early 2000s fondly as being a time where you could discover an operating system that was new or novel in some way. Combine that with the fact that computers felt less laggy, and mostly used the same UI across applications (looking at you electron), and I share the nostalgia.
I guess what I’m saying is that an OSS OS/2 is cool, and its authors should have songs sung about them, but this isn’t the right foundation to build a modern OS on.
It lent itself to being tinkered and tailored in a much more developer-friendly way than most OSes. The rexx scripting was pervasive. There were meaningful file object attributes that could be extended and used per app. The SOM model was a bit clunky but very powerful (I went on to work with its designer later in my career). It had a bunch of powerful compilers that left all the Windows competitors for dust.
And of course it looked great for a windowed GUI at the time.
Scripting and generic IPC hooks are one of the coolest things computers can do, and they're more and more limited as time goes on. Just this morning I tried writing a Dark Mode menu bar app for OSX (I'm aware of NightOwl) and it's literally 5 lines of code, but I can't get it running because of sandbox headaches. Looked up all the combinations of solutions but couldn't figure it out after 30 minutes. All sorts of automation and inter-app use-cases are basically dead. All in the name of greater security, except that it can still be done with a good security model. It's possible, it's just that that's not anybody's use-case for computers anymore, except for power users who know how to use Linux, so it's been downgraded to "unsupported". (And Linux is just generally not compatible with rMBP hardware afaik.) Reading what you said about OS/2 sounds like it was a lot closer to this ideal.
Rexx scripting was great! And it definitely felt like they were catering the OS to a powerful group of power users.
To clarify my statement above about its code being a hot mess: my understanding was always that OS/2 achieved reliability through sheer force of will rather than it being built in a way that was architected to be intrinsically safe. Sort of how US nuclear subs have achieved an impressive safety record mainly due to extraordinary fastidiousness with respect to process and procedure.
When did you develop for it? I remember 89 to 95 being the heyday of OS/2, when if you wanted to get your Windows 3.1 machine on the network, you pretty much had to use OS/2 or Novell to host the network.
This was approx 1990-1998. I was working for an actuarial partnership. Our software generated customer-specific actuarial FORTRAN code, with the generator and UI in C. We used the Watcom compilers for both ends of the pipeline. They were top class, and still available AIUI.
I've only played a bit with eComStation (recent OS/2 "distribution"), but my understanding of OS/2 is that it is a very modular system which is also how people still keep adding features to it despite using the same base since the 90s. Also it seems to be thoroughly documented.
I think both of these, combined with the fact that OS/2 is not a moving target anymore (unlike Windows), can make it much easier to make an open source OS/2 replacement by replacing its systems bit by bit. If anything, if you check eComStation (and i also guess Arca OS), it has a lot of newer parts than what was originally in OS/2 at all stages of the "stack" - from the bootloader, to graphics drivers, to UI, many of which seem to be open source.
TBH i think the main reason there haven't been already a working OS/2 replacement is that there doesn't seem to be enough of a technical know-how in the current community and there doesn't seem to be any new interest for it by people.
Haiku was pretty much in the same position, but the difference that somehow Haiku managed to get enough knowledgeable people on board despite BeOS being released on PC only a couple of years after OS/2 Warp. I guess the release of BeOS Personal for free helped spread it much further and OS/2 never had something like that.
Well, it's rather a OS/2 look-alike, i.e. compatible OS, rather than a reverse engineered re-implementation.
Based on other free software packages, e.g. the interesting Fiasko (from the L4 family) micro kernel, one could rather see it as an example application of those.
This brings back memories! I was part of TVT (translation and verification team) for os/2 in 1996. Still remember how developers were supplying fresh builds to install and verify. We were a bunch of people from all over the world trying to do our best and localize os/2, but you could hear the boat sinking - windows was just too good - which is a pitty since from technical standpoint os/2 was better.
I have to check I sill might have some cds from that era...
> windows was just too good - which is a pitty since from technical standpoint os/2 was better.
I grumbled about this through a fair part of the 1990s. I ran Warp for a while and programmed for it as well as for Windows 3.1 and Windows NT 3.5; on the machine I had at the time there were Windows 3.1 programs that actually ran better on OS/2's compatibility layer than they did in native Windows.
True. One thing I remember is that it had shortcuts which were introduced only in win95. Unfortunately its UI was not very intuitive.
Not to mention one had to pay for it when win 3.1 was widely available for free at your local pirate.
> I have to check I sill might have some cds from that era...
I know that the Internet Archive (https://archive.org/) are always willing to host CD images for their software collection. They accept uploads from anyone, but Jason Scott (@textfiles on twitter and HN) is their software curator if you need assistance.
Ah OS/2 Warp. The OS I got grounded over because I reformatted the family 486 and lost some of my parents files. I truly thought I was setting up dual boot. I was so naive at the time.
When I was in university (Fraunhofer, actually) we built a optical metrology machine (CMM) and the SW was based on OS/2 1.2, which had reliable threading and was generally rock solid compared to Windows. Linux was not a thing yet.
After graduating I moved to the US and got my first job due to my OS/2 experience. The company I worked for had a Mac/PC network bridge product (TCP was not a thing on PCs yet) and we were contracted by IBM to write LAN Server for Macintosh, basically the Mac/AppleTalk server for OS/2. Good memories.
As someone who supported software and Mac/PC networks in the 90s, and someone who has done frontend development since 1995, I have a keen appreciation of the irony here.
It would be interesting to see a Wine-like compatibility layer for OS/2 applications. It could even be included in Wine itself, seeing as there are quite a few analogies between Win16/Win32 and OS/2.
OS/2 always intrigued me because it looked like a strange cousin of Windows; it so clearly shared the design language (and some actual code) with that family of operating systems.
I recall using GEM on the Atari ST (the first computer I bought, for about $50 at a garage sale), and then using Owen's FreeGEM distribution at some point later on. I didn't really think it fit in with the theme of being DOS-Windows-OS/2 ish, although the Atari ST veresion seemed to use some modified form of FAT on floppy disks (based on the fact that the filesystem was readable but with mutations on the family DOS/WFW 3.11 machine). I certainly didn't feel like Windows in the same way that OS/2 feels like Windows.
Well, the Windows NT family (which includes Windows 2000, Windows XP, Windows 7, Windows 10) is a descendant of OS/2. In fact, Windows NT 3.1 was basically a reskinned OS/2.
I remember fond days of playing several MSDOS games at once under OS/2 (Elite: Frontier and Colonization, iirc) while Windows 95 failed to play any of them :)
This isn't quite true. Microsoft and IBM were working on OS/2 together and Microsoft got fed up with the development delays and decided to build their own system. So NT was designed from the ground up. It originally had multiple subsystems (Win16, Win32, Posix) one of which included an OS/2 1.0 subsystem.
With WSL bringing back the spirit of a POSIX-flavored subsystem, it might be fun to see some other retro-computing subsystems such as a modern OS/2 subsystem, though I can't think of a good reason to do that other than fun.
Maybe if someone can think of a good reason to need OS/2 running on Azure it could happen. Maybe there's a bank somewhere with the OS/2 apps investment and enough money to throw at Microsoft (and IBM) somewhere to make it happen.
> Maybe if someone can think of a good reason to need OS/2 running on Azure it could happen.
FYI you can do this today with VMware either on your own hardware, or within AWS, running ArcaOS 5 (the successor to eComStation, i.e. the maintained/supported flavor of OS/2 that lives on). Though I would be shocked if anyone has actually tried it:
WSL2 is certainly doing fewer kernel interactions as a subsystem than WSL1, but my impression is that it is still classified as a subsystem in the way that it launches and manages its more VM-like approach.
The comparison/analogy I've heard, and I'm not sure it is apocryphal but it certainly makes sense to me, is to the Xbox One's game subsystems, which while rarely described in great detail from an NT Kernel engineering standpoint, when they are briefly described as such they are often described as subsystems and that they are "VM-like" in operation.
Yes, I mentioned all of that. All I was saying is that WSL2 is still technically a subsystem by name, even if it is "just running the Linux kernel". NT Subsystems have always varied in nature of how they operate. Yes, it's disappointing that WSL2 is "less cool" or "less proper" a subsystem, but they still call it a "subsystem for Linux", not a "Windows VM Host for Linux" or anything like that. (It might be less confusing between WSL1 and WSL2 if they had had chosen a name like "Windows VM Host for Linux" for WSL2.) To the NT Kernel naming perspective these "micro-VM hosts" still appear to be called "subsystem", even if they are doing so much directly less Kernel work than a "proper" subsystem in the Win16/Win32/Classic POSIX/OS/2 1.x or even WSL1 sense.
Though this is getting into semantic weeds and drifting away from the original point that more subsystems, especially retro-computing inspired ones would be fun to see, even if they are more just "micro VM hosts" than full NT Kernel "flavors".
(ETA: Also, it's a small stretch that Microsoft "gave up" on the WSL1 approach. They will still be supporting WSL1 side-by-side WSL2. The numbered name approach makes it seem that it is a strict "upgrade", but it's more a side-by-side trade-off option than the numbers imply.)
The fact that if you look at the full callstack and address space of a regular Linux process running in WSL2 you won't see the "classic" NT Kernel anywhere feels like a pretty big difference to me.
Yes they are still supporting WSL1. Microsoft's support policies are generally excellent. But it seems very unlikely there will be an interest in providing OS emulation or compatibility at the userspace/OS kernel boundary using the classic subsystem architecture, as a technical direction for the Windows team to invest in going forward. It's an architectural dead-end.
There is always an 100 different stores when corporations break up.
The reality seems to be that MS kind of wanted something more lightweight that could be sold to existing pc owners, where as IBM was happy to just push the requirements to the point where OS/2 would require an fairly high end system(preferably an PS/2 from IBM to run).
For some markets where the expensive PS/2 systems were not considered too expensive as in the case of banks and the like OS/2 sold well enough for IBM to profit off it. But the wider market went with the less hardware hungry and at least initially less capable windows.
If i recall correctly NT was slow to mature into an mainstream OS and did not really start replacing OS/2 until the late 90ies and only really took of for consumers when XP were released.
I see a lot of people forgetting Microsoft's illegal "white box tax" as the years go on. Anticompetitive practices meant many small businesses and home users were seldom going to use anything that cost them money on top of the DOS or Windows they already had to pay for just to own a computer.
Kernel-wise, there isn't much common code between OS/2 and NT. Windows NT 3.1 is more like VMS with the Windows graphical subsystem ported on top of it.
But yes, running OS/2 was fun days, much better than Windows 3.11 (if you had memory enough).
Sorry to be pedantic, but Windows NT 3.5 was the VMS version of windows to compete with OS/2, and Windows 3.0 and 3.1 was the mass market version built on top of DOS. If you bought through a small shop, they would add a batch file to autorun the windows executable that launched the GUI from an ini file. Nobody had the Windows 3.11 version if I remember correctly
I don't think you read my comment. I know that Windows 3.x were not based on NT. They were DOS shell extensions essential that could run DOS, but not as well as OS/2. Windows 3.11 didn't sell hardly at all, but everyone mentions it like it did. It is a common mistake.
Windows NT 3.1 was just something I forgot about. Windows 3.51 was the popular version. All this was before the internet as we knew it took off. People only had access to what was available in the book store or computer store, and online updates was usually for enterprise subscribers only.
Yep, the schism between IBM and Microsoft let to the two products. I preferred OS/2 and workplace shell at the time. Heck, I was using OS2 up until 2000ish. Then got sucked into the land of BSD.
I still have a vm floating around of OS/2 Warp so I could play the original Galactic Civilizations from Stardock
Please write about some of these UI features, I would love to hear about it! I never got to play with it, because it demanded a much better computer than I had.
I think the GP was referring to features that were advanced for the day, but these days we take for granted, like a system control panel controlled entirely through a GUI rather than any text files like Autoexec.bat and config.sys
Also features that we should be able to take for granted but are broken or missing in current systems, such as spatial file managers.
In OS/2 and classic Mac OS, folders you opened would remember their window position and other properties (including custom background image in OS/2) so that you could find things in a very human way by remembering e.g. "the small blue window in upper left".
Windows 95 copied this in a slightly broken way – you could open the same folder in multiple windows at the same time and give them conflicting properties, which breaks the model. Windows 98 (and the IE 4 upgrade to Windows 95) merged the whole thing with the web browser which broke it even more.
Mac OS X also had a half-broken implementation of this, since the file manager tried to combine the Mac OS Finder with the NeXTSTEP file manager which was completely different. John Siracusa has written extensively on this topic: https://arstechnica.com/gadgets/2003/04/finder/
However, UWP now makes COM quite nice to use, this was what VB 6 and C++ should have turned into, instead of racing after Java, and taking 20 years to do a full circle.
According to the whitepaper, one of the (original) goals is/was to create various "personalities", including that of Linux, OS/2 being the "base" personality.
They mention their own bootloader called freeldr. Interestingly, this is the same name as ReactOS’s bootloader. I assume it’s just a coincidence, but since I know OS/2 and NT share some common lineage, I do wonder if there’s any relation. A bit of searching didn’t seem to clear it up. Anyone have a clue?
It's a long time since I do not follow this topic, but I think Freeldr is (was?) an independent project. I think also it was introduced in Reactos at the end of the 0.2.x branch.
For example it was not used in Captive Reactos and pre-2005 versions.
There was something cool and sweet about the MS-DOS command prompt simplicity (compared to UNIX) and the OS itself being single-user with a (limited) multi-tasking capability. I have always dreamed of a true 32-bit continuation of it, and OS/2 seemed to have promised to be that, but something happened along the way...
The early versions of Citrix were based on OS/2, we used it to run a FoxPro based DOS app over our WAN for quite a few years. For the time it was a pretty innovative/ performant way to run apps over low bandwidth.
The largest and oldest bank here in Brazil migrated all of its ATMs and branch manager desktops from OS/2 to Linux a decade ago; the other large banks I know of in Brazil seem to use Windows for desktops (and probably also their ATMs). I don't know about other countries, but it wouldn't surprise me if their banks have also migrated to something other than OS/2.
I was at the help desk for an large Scandinavian bank doing the late 00ies and they were running virtual OS/2 instances on nearly every windows pc in order to support an custom app, and still had a few front desks(or whatever the window where the bank teller sits is called) with dedicated OS/2 hardware.
The biggest problem here was sourcing dot matrix printers as the software used as the front desk software only worked with a few models of no longer in production printers.
A few years after i left i heard rumors of an transformation project to get rid the legacy interfaces, part of which had been on going for a decade(as an old school java application) by the time i left for greener pasture, but i would not be surprised if that project it is still not complete yet, nor would i be to hear of retail banks that never bothered with any investment to replace OS/2 based apps.
Pretty old. I want to say I built their website back in 2008 or so (they changed the logo and color scheme but it seems to still use dokuwiki and the base templates I built out), and it was definitely going on longer than that.
Probably won't happen. Companies have been willing to buy the IP and support it (Serenity Systems with eComStation, Arca Noae with ArcaOS) because there are legacy installations of OS/2 that need to be supported (and are willing to pay). To get a sense of the companies that still need OS/2 support, see this list from Arca Noae:
Advantech, Austria
Agie Charmilles SA, Switzerland
Bowe, Bell & Howell, U.S.A.
Canadian Coast Guard, Canada
Colgate Palmolive, U.S.A.
Dematic North America, Luxembourg
Federated Insurance, U.S.A.
Johnson and Johnson, U.S.A.
Pepsi Co., U.S.A.
Roos Instruments, U.S.A.
Siemens AG, Germany
Trainitalie, F.S., Italy
Universal Instruments Corp, U.S.A.
Whirlpool Corporation, U.S,A,
One of the reasons why the original OS/2 source wasn't made available by IBM was that there are a lot of 3rd party libraries in it. Many from companies that no longer exist, or firms that have been acquired several times, so it's just too hard to get the release signatures.
That was true for IBM mainframes manufactured prior to 2005. The hardware management console computer (kind of like BIOS setup on steroids) was a PC running OS/2. Models since 2005 have used Linux instead, although the management app on Linux was designed to duplicate the visual look of OS/2 (to make existing users feel more comfortable.)
I remember around 2006 or so, watching a service technician replace some faulty hard disk drives in an enterprise SAN. He opens up the front door, and out of this compartment he slides a laptop computer which runs some management software. From the look of the user interface, it appeared to be running Windows 3.1.
Raises hand I have a Multiprise 3000 and this is true for me, or would be if I could get the software that will actually bring up the mainframe hardware.
OS/2 was tied to the PS/2 hardware and IBM had it way over priced. We had one in the office and when we needed a new card for a peripheral, we realized it was never going to cut it compared to compatibles.
OS/2 was not tied to the PS/2 hardware. Unless there was some specific version of it. It was reasonably compatible with a fairly wide range of hardware. (Roughly comparable to Linux desktop at various times)
Maybe I'm remembering "guilt by association". IBM was pushing the PS/2 line very hard because they wanted the compatible hardware to die and they assumed everyone would still trust them.
I don't remember ever trying to install OS/2 anywhere else.
OS/2 was developed from scratch, and was not encumbered by a legacy of DOS compatibility, 16-bit applications, and drivers like Windows was. This enabled IBM to innovate radically and more quickly. The application I wrote for my client was a multi-threaded program, written paradoxically with Microsoft's Programmer Workbench, the character-based precursor to Visual Studio.
However, this freedom from legacy code also meant there were few applications available for OS/2, and the only drivers were for IBM printers. OS/2 eventually disappeared in spite of its technological superiority at the time.
IBM's development environment was called C-Set, and, despite being a 32-bit compiler, was inferior to Programmer's Workbench. For instance, after a compilation, if you double-clicked on an error message, it would open up the source code in a separate editor window, in a proportional typeface. You could not trigger a compilation from this window. You had to make your change, save & close, then go back to the compiler window. Awkward.
There was no Resource Workshop for designing dialog boxes visually. You had to write all the statements in a text editor, and wouldn't see the appearance until compiling and running the program. This slowed down development.
While debugging, I recall trying to step into the GUI part of the application code. This locked up the computer so bad we had to re-install OS/2, a tedious process requiring 22 diskettes.
The IBM documentation was so inscrutable that even for ver. 2 of OS/2, we had to refer to Microsoft's ver. 1 manuals. IBM's support gave an interesting insight into OS/2's downfall. My client had an expensive OS/2 support contract with IBM. The client contacts were two guys in the PC support section. When I required technical support to answer some obscure and specialized questions about OS/2, it didn't make sense to get these guys involved, even though they were supposed to be the only official go-between with IBM. So I called IBM with the client's authorization, and impersonated one of the support techs. After a few calls I slipped and used my name. The IBM guy was livid and tore a strip of me, saying that as an independent contractor, I had no business blah blah. Here I was just trying to get a customer up & running.