OS/2 vs. early Windows

[lowen]

My notes say that SLS was announced in August 1992. While was X11 likely was included, I'm not sure how useful it was at that point, I have not used it. Also, I did see that the network stack and drivers were in a bit of a flux. Also, SLS was considered buggy. So as a 32-bit OS in 1992, it was not quite ready.

I ran SLS for a short time a month or so before I met my wife, and I met her on August 14, 1992. But SLS was not the only 1993 Linux distribution. See https://lunduke.substack.com/p/the-linux-distributions-of-1992 for more, including TAMU, which also had XFree86 included.

Yes, SLS was considered buggy, and thus Debian and Slackware were conceived; both are still updated to this day, and I'm personally running Debian 12.

 

[lowen]

I ran SLS for a short time a month or so before I met my wife, and I met her on August 14, 1992.

Correction: in looking back in my email storage from back in the day, I downloaded SLS after I met my wife, not before. Downloaded from Mac's Place BBS over 9600 bps dialup. Looking in Usenet archives, Peter announced SLS as being available for download 8/15/1992; the May 1992 date is referred to all over the Internet as the date it was 'founded' whatever that is supposed to mean.

 

[Makefile]

Correction: in looking back in my email storage from back in the day, I downloaded SLS after I met my wife, not before. Downloaded from Mac's Place BBS over 9600 bps dialup. Looking in Usenet archives, Peter announced SLS as being available for download 8/15/1992; the May 1992 date is referred to all over the Internet as the date it was 'founded' whatever that is supposed to mean.

Did you come to Linux by way of Minix or straight from Tandy Xenix?

 

[lowen]

Did you come to Linux by way of Minix or straight from Tandy Xenix?

Xenix -> AT&T/Convergent System V Release 2 on 3B1 -> Apollo DomainOS on a pair of DN3500's -> Linux

There were a few detours along the way, like a short time with straight MS-DOS plus Waffle on an Epson 8088, a few months with DESQview on a 386SX with 5MB, a short run with Windows 95 (on that same 386SX with 5MB of RAM), but by the time Red Hat Linux 4.0 was available I bought a boxed set and ran RHL. I've been running a Linux of some kind as my primary desktop ever since.

That's my personal desktop; obviously I've had to use Windows or whatever at work, but in the current day job I'm running Linux, with Windows of various versions in VM's when required.

More to the topic at hand, I did evaluate OS/2 back in the day, and it really was solid. It would have been a much better system, in my opinion, than the thunking system of VXDs we got, known as Windows/386 (the VMM386.VXD operating system's lineage from Windows/386 2.0 through Windows 3.0 enhanced mode through Windows for Workgroups 3.11 where VMM386.VXD with its various thunked-in 32-bit drivers really became its own OS, only using DOS as a loader through Windows 95 (where DOS as a loader became 'hidden') through Windows 98 and ending with Windows ME -- the VMM386.VXD OS grew organically).

We did finally get what OS/2 promised with Windows NT.

But my personal wish would have been a Xenix core with a reasonable GUI on top, but that didn't happen.

 

[MennoJoshua]

Win 3.1 has the advantage it could run acceptable in 2MB of RAM and performed well with 4. It also was usable in a 286 and performed well on a 386.

MS OS/2 2.0 actually performed pretty well on a 4MB 386 but unfortunately never got released. It predated Windows 3.0!

 

[soviet9922]

Other simple things that where a pain is for example you installed a LAN driver or something and OS/2 stop booting, you where kinda fucked removing it could be complex. On windows 95 at least you have the "safe" mode that allowed you to boot and remove the device/program from the GUI of course it was far from perfect. But this kind of usability things where hard on new users.

 

[vwestlife]

Other simple things that where a pain is for example you installed a LAN driver or something and OS/2 stop booting, you where kinda fucked removing it could be complex. On windows 95 at least you have the "safe" mode that allowed you to boot and remove the device/program from the GUI of course it was far from perfect. But this kind of usability things where hard on new users.

OS/2 Warp's equivalent of Safe Mode is recovery mode, accessed by pressing Alt-F1 when the OS first starts booting.

PSM: OS/2 Warp - OS2World.Com Wiki

www.os2world.com

In OS/2 2.x you can boot from the installation diskette and then access a command prompt.

 

[MiaM]

The boot menu in DOS 6.x must partially had been the result of frustration from faulty configuration requiring a DOS boot diskette. Or rather perhaps not the boot menu itself, but the hold-shift-to-bypass-config.sys-and-autoexec.bat feature.

Interestingly DOS 6 seems to have been released two months before OS/2 2.1 (but the year after OS/2 2.0). This kind of boot menu would likely had been an attractive feature to OS/2, but then a pure single user text mode (probably using BIOS text modes) would had been required, and I don't know if OS/2 2.x supported that?

 

[traveler_aa5]

This is from an OS/2 VM running under VirtualBox on Linux. I had to drop the execution cap to 4% so I
had enough time to press ALT-F1.

OS/2 could boot to a command line. Press "ALT-F1" when booting, and you had several options

https://imgur.com/a/laULB52

If you select F4, it gives you a GUI.

https://imgur.com/a/mfRQLyv

 

[bear]

The boot menu in DOS 6.x must partially had been the result of frustration from faulty configuration requiring a DOS boot diskette. Or rather perhaps not the boot menu itself, but the hold-shift-to-bypass-config.sys-and-autoexec.bat feature.

It was more because of the difficulty getting enough conventional memory free for a growing number of programs that wanted more of it than people were ending up with, after loading all the device drivers. With or without network drivers. With or without EMS. Set up for best operation of Windows for Workgroups; set up for best operation of DOS. These things all existed in various states of conflict with one another. If you needed both Windows, and (say) access to a Novell network for a DOS program that ran poorly (or not at all) in Windows, you could easily end up losing to the Novell stuff 100KB+ of conventional memory that you would badly want back when you _weren't_ using that program. Other folks used it to set up tuned environments for DOS gaming, especially for games that supported network multiplayer, and trebly so if you had two games that both relied on different network stacks. They were not great times.

 

[CaptainWillStarblazer]

I feel like you guys are overstating exactly how useful Linux was in the early days. While a handful of people did use it, the hardware and software support (particularly software - almost nothing outside of freeware) was incredibly poor. You could forget about playing games or running decent productivity software, and it was just not a viable alternative to Windows or OS/2 or anything else.

To OP's point, OS/2 32-bit was not released too early. In fact, Microsoft wanted it to be 32-bit from day one, and it was down to IBM that OS/2 originally targeted the 80286. By 1992, most new computers were 80386 machines and they had been steadily getting more and more common for the last few years (despite being extraordinarily expensive from 1986 into the late 80s, prices were coming down in the early 90s) so there were a lot out there. The 80286 was just a technological dead-end. IBM held OS/2 back unnecessarily.

With that said, the "marketing can beat real capabilities" refrain about how OS/2 should have beat Windows is just overstated. While non-NT Windows was far from solid, it at least had a large library of applications, something that OS/2 could not claim, and ran well on affordable machines of the day. Moreover, from a technical point-of-view, Windows 3.0 blew OS/2 1.3 (its contemporary) out of the water due to Windows 3.0 supporting multitasking of both protected-mode Windows apps, protected-mode DOS apps, and real-mode DOS apps all at once, and only for the brief period between the release of OS/2 2.0 (1992) and Windows 95 (1995) did IBM have the technical lead. Windows 95 had multitasking that was just as good plus a way better software library, and better APIs for multimedia that allowed games to move from DOS to Windows.

The few people who did use OS/2 were very committed fans who would rave about it to anyone who would listen (not unlike Amiga fans today, or Linux users); they loved it, but even they couldn't deny flaws like the SIQ.

 

[CaptainWillStarblazer]

Presentation Manager was solely an IBM effort. The single input queue in PM was also an IBM design. IBM chose to rely on PS/2 specific features instead of something that worked well on AT and PS/2 machines.

Presentation Manager was not an IBM-only effort. In fact, as with most of OS/2 before the JDA breakup, Microsoft developed most of it. The entire design of the Presentation Manager is clearly based on Windows (both in look and feel and in API design - while GPI is clearly inspired by IBM's mainframes, the window management component of PM is almost copied from Windows USER), just with some CUA sprinkles added in (that filtered back into Windows, as they were supposed to share a UI to allow people to transition from Windows to OS/2, the "OS of the future"). After PM started, Windows and OS/2 had their UIs developed in lockstep and there was always supposed to be some synergy between them. Presentation Manager is better thought of as IBM taking a look at Windows and giving Microsoft line notes on what to change, rather than a purely IBM effort.

Moreover, what PS/2-specific features in particular are you referring to? IBM might have had some nice PS/2 specific code paths for optimizing stuff like toggling the A20 line, but they certainly supported generic AT machines too. IBM definitely put a priority on drivers for IBM hardware (especially video and network adapters), but the video was cloned widely enough for that not to be a major issue.

 

[ajacocks]

I feel like you guys are overstating exactly how useful Linux was in the early days. While a handful of people did use it, the hardware and software support (particularly software - almost nothing outside of freeware) was incredibly poor. You could forget about playing games or running decent productivity software, and it was just not a viable alternative to Windows or OS/2 or anything else.,

I have to take exception to that statement. "Freeware" is not an apt description for open source and free software, to begin with. The US gov't had a whole long treatise on not using shareware and freeware, but specifically called out FOSS (free and open source software) as being different, quite a long time ago. As someone who has been a Linux user since 1992 (not day 1, or anything like it, but SLS early), I can say with authority that Linux was indeed useful not long after its first availability. SLS (Softlanding Linux System) was quite usable, if not bug-free, and the release of Slackware in 1993 took care of that problem. It was indeed a viable alternative to Windows or OS/2, depending on what your objectives were. As a computer science student, at the time, and as a US government employee (NIH, gods rest its soul!), I used Linux daily to do my work, and it was a fantastic solution.

I will give you that commercial software availability was not huge, but the evolution of computing since then has shown that FOSS is the driver for innovation, and that *NIX rules the world. See market share statistics here, for example.

This does not at all mean that I disliked OS/2. I did use it then, as well, and liked the versions starting with 2.1. I purchased and used 2.1, Warp 3, and Warp Connect to do a lot of desktop usage. It is indeed too bad that the IBM/Microsoft relationship went the way that it did, since at the time, OS/2 was definitely a strong and reliable desktop, as seen by its extensive use in embedded solutions like ATMs.

Disclaimer: I work for Red Hat, and have spent my entire working life supporting FOSS on many platforms, both open and closed.

- Alex

 

[vwestlife]

Presentation Manager was solely an IBM effort. The single input queue in PM was also an IBM design. IBM chose to rely on PS/2 specific features instead of something that worked well on AT and PS/2 machines.

AFAIK, all releases of OS/2 1.x were OEM-specific. There was no "generic clone" version of it. The closest you could get was an Intel 386 OEM version of Microsoft OS/2 1.x. I know, because I have a boxed copy of it.

 

[krebizfan]

AFAIK, all releases of OS/2 1.x were OEM-specific. There was no "generic clone" version of it. The closest you could get was an Intel 386 OEM version of Microsoft OS/2 1.x. I know, because I have a boxed copy of it.

IIRC, the OEM versions were based on the OS/2 for the IBM AT with only minor patches for those that weren't 100% AT compatible.

MCA based systems used the watchdog timer to help prevent lockups. I don't know if any version of OS/2 1.x was updated to use the EISA watchdog timer.

 

[CaptainWillStarblazer]

AFAIK, all releases of OS/2 1.x were OEM-specific. There was no "generic clone" version of it. The closest you could get was an Intel 386 OEM version of Microsoft OS/2 1.x. I know, because I have a boxed copy of it.

To your knowledge, were the OEM versions of OS/2 actually customized in major ways (i.e. besides adding device drivers), or were they basically just branded, like MS-DOS? To my knowledge, even though there was no "generic" unbranded MS-DOS for a while, all 100% accurate PC clones (i.e. Compaqs and stuff) ran basically identical code.

I have to take exception to that statement. "Freeware" is not an apt description for open source and free software, to begin with. The US gov't had a whole long treatise on not using shareware and freeware, but specifically called out FOSS (free and open source software) as being different, quite a long time ago. As someone who has been a Linux user since 1992 (not day 1, or anything like it, but SLS early), I can say with authority that Linux was indeed useful not long after its first availability. SLS (Softlanding Linux System) was quite usable, if not bug-free, and the release of Slackware in 1993 took care of that problem. It was indeed a viable alternative to Windows or OS/2, depending on what your objectives were. As a computer science student, at the time, and as a US government employee (NIH, gods rest its soul!), I used Linux daily to do my work, and it was a fantastic solution.

I will give you that commercial software availability was not huge, but the evolution of computing since then has shown that FOSS is the driver for innovation, and that *NIX rules the world. See market share statistics here, for example.

This does not at all mean that I disliked OS/2. I did use it then, as well, and liked the versions starting with 2.1. I purchased and used 2.1, Warp 3, and Warp Connect to do a lot of desktop usage. It is indeed too bad that the IBM/Microsoft relationship went the way that it did, since at the time, OS/2 was definitely a strong and reliable desktop, as seen by its extensive use in embedded solutions like ATMs.

Disclaimer: I work for Red Hat, and have spent my entire working life supporting FOSS on many platforms, both open and closed.

- Alex

Your statement that "FOSS is the driver for innovation" is debatable at best given that Linux is essentially just a clone of an OS design from the 70s that had been old and tired for 20 years, while the design of NT was far more modern. Moreover, stating that *NIX rules the world is reductive at best. I look at your graph and see Windows as the top desktop operating system. Sure, Android is on more devices in total, but it's not competing with Windows, they're in a totally different segment. Similarly, Linux is dominant in servers (though Windows seems to have carved out a niche in corporate intranets), but that's also a totally different market. In the computers on most people's desks, it is Windows that powers them.

It is most definitely not "too bad" that the IBM/Microsoft relationship went the way that it did, as IBM was dead weight in the relationship and refocusing on Windows (and shifting NT to be Windows NT) ended up being the right choice for the market.

To your point on "freeware", I sometimes get irritated at the semantic pedantry of people in the FOSS world, but I'll admit that I was in error here. With that said, it is still true that basically your only options on Linux at the time were to use FOSS. Was Linux useful for some people on day one? Sure, but you had to be very clear on your needs or prepared to compile a whole lot yourself. Even a lot of developers don't want to have to compile a bunch of other people's code for every application they run. I've yet to find an IDE or debugger on Linux that matches up to what I've found on Windows. The open-source office suites have left a lot to be desired compared to Microsoft's, and while Linux gaming is now excellent, it was in a pretty rough state for a very long time.

 

[ajacocks]

You have plenty of fair points, @CaptainWillStarblazer. Linux definitely isn’t the end-all and be-all for FOSS, however. It’s certainly well-known, but plenty of projects predate the initial kernel release. Microsoft itself seems to have a belief in FOSS, based on this and their many community contributions.

On the NT innovation side, yes, Dave Cutler and team had some fantastic ideas. However, the front end put on that team’s kernel and internals does not in any way reflect the underlying architecture.

 

[CaptainWillStarblazer]

You have plenty of fair points, @CaptainWillStarblazer. Linux definitely isn’t the end-all and be-all for FOSS, however. It’s certainly well-known, but plenty of projects predate the initial kernel release. Microsoft itself seems to have a belief in FOSS, based on this and their many community contributions.

On the NT innovation side, yes, Dave Cutler and team had some fantastic ideas. However, the front end put on that team’s kernel and internals does not in any way reflect the underlying architecture.

It is my honest opinion that NT is the only mass-market operating system that embodies a large amount of new good ideas. Personally, I actually quite like the Win32 front-end and find that it is very well-designed. While I question most of Microsoft's choices in the last 10-15 years, I generally like a lot of their design.

Linux and BSD are not great sources of innovation, and while no doubt there is some great open-source work going on, none of it has hit the same mass appeal in the FOSS world as Linux or BSD. I don't inherently fault Linux for this, as it does what it originally set out to do; it is a free UNIX for PCs, but I just don't think that's a great goal to aspire to. But in spite of whether or not you think Linux has grown to become something great now, you can't deny its severe flaws in usability at the time. Was it the start of something great? Technologically, I don't think so, but culturally, it absolutely is. The FSF and the movement it embodied existed before Linux, but it took a free kernel (that the GNU project famously failed to deliver, and that BSD had been legally encumbered in its attempt to do) for that movement to really take off, and while Linux could do a lot of cool things (I've not heard good things about SLS but Yggrasil Linux seemed neat as least as a showcase), the movement was very much in its infancy and the limitations were obvious. Were some enthusiasts willing to deal with that burden? Of course. Early adopters are always willing to deal with flawed products. But it was still a flawed product.

 

[MiaM]

Re Linux (and *BSD) v.s. OS/2:
I would say that at the time Linux (and *BSD) actually became usable for at least "major nice applications" (I.E for example a web server) the OS/2 v.s. Windows war was already over and OS/2 had already lost. IBM kept releasing OS/2 Warp 3 and 4 even when they had already lost.

As already stated, the software library for Linux (and *BSD) at the time of the actual Windows v.s. OS/2 "war" kind of only contained development and OS related stuff. I.E. it was great for someone wanting to develop UNIX software, or in general just wanted to develop software for any "unspecified" OS that would multi task and be more stable than most alternatives, while not costing an arm and a leg. It might also had been usable as a terminal (X11 and text mode) to other UNIX systems.

Before Windows 3.0 there were "no" software for Windows. Or rather, there were Excel and Word, and a few other applications that didn't have any large user base.
I know that WinWorldPC obviously only lists software that actually exists on the site, but in general if you do a search for Windows and date up to 1989 there are 54 search results:
https://winworldpc.com/search?showF...=1989&vendor=&platforms=Windows&sort=alpha-az
Searching for OS/2 up to 1989 yields 20 results:
https://winworldpc.com/search/?show...Year=1989&vendor=&platforms=OS2&sort=alpha-az
While the library for Windows before 3.0 was more comprehensive than for OS/2, the difference doesn't seem that large.

 

[krebizfan]

The figures I have indicate about a million copies of Windows 2 were sold in 1988. The estimated total of Macs sold was 1.27 million in that same year.

Additionally, Windows applications could be sold with only a runtime. Effectively, most PC copies of PageMaker for Windows and Excel for Windows were stand alone DOS applications. Windows applications outsold all other GUI applications (except Mac) by a lot even before Windows 3 made it to market.