SAA is what OS/2's presentation manager was built around. Some pipe dream that the mainframes, as/400's and rs/6000's were going to share a common UI. Well that never happened, and it was a kneejerk thing to push MS out of the UI on OS/2.
Not that it matters, MS-DOS had a lot more device drivers than OS/2, and Windows 3.0's ability to use them made it a winner.
OS/2 was more of a learning tool in how not to push people off of MS-DOS, and instead they moved to Windows, then once machines were fast enough and ME was horrible enough, everyone went to XP Home, and plenty of users are still there.
They kind of did with LanMan server, and things like SQL. The OS/2 Extended Edition bundled lots of stuff together, but it was IBM's way of doing things, and I never saw anyone using EE. However I've setup MS SQL 1.0 on OS/2 and it is a NIGHTMARE. Compared to NT where you install NT, then SQL and away you go. But no, Install OS/2, reboot install the lan driver stuff, reboot, install lan man server, reboot make sure you can now create named pipes, and read them, then install sql server. OS/2 refused to bundle in the important bits, that NT and WfW later would all bring in by default. It didn't take a genius to see the rise of the LAN, however it took some major pushes to get into server space. But nobody enjoyed dealing with netware and their NLM crap, once NT hit v 4.0 everyone was dumping it for NT. But the underpinning network aspects of NT were on OS/2 + LanMan. NT just started from that point and had it all built in.
What is weird is how MS saw the LAN, but missed the internet. Even OS/2 3.0 Warp with that woefully useless IAK, provided no LAN access, or any peer to peer networking capabilities.
Granted the window between 3.0 and 3.1 for workgroups was a crappy period for networking but by the time OS/2 3.0 came out, it was already too late. And it's TCP/IP was that horrible dialup centric POS, nothing multi-homed, or with physical network cards.
in the OS/2 1.x there was a clear leader in the LAN, and it was Netware. But IBM just thought OS/2 should be as crappy as MS-DOS, and let the 3rd parties come along and add in support. Instead they should have at least bundled in LanMan support, but of course that would have killed a part number.
Windows for Workgroups 3.1 did have a network stack, and it was greatly improved in Windows for Workgroups 3.11. NT 3.1 came with TCP/IP, IPX, and support for SNA out of the box in 93. Warp was in '94.
The real 'killer' of course was Win32s+Winsock which gave us Mosaic. And why was IBM not killing themselves to make a Mosaic port to OS/2? Once people saw a graphical Internet, everyone I saw around me was clamouring for it, and it was amazing at the time how many Windows 3.1+Win32s+Mosiac installs talking to various UNIX dialup accounts using SLiRP. Something that was really unstable on OS/2, so more of the OS/2 fans were either going to Linux or to NT 3.5/3.51
It wouldn't matter, as SMP was becoming a thing, and don't forget the coming x86_64 along with the ability to run on RISC. OS/2's kernel was largely untouched from early MS OS/2 2.0 betas, and the device drivers were still 16bit assembly. IBM's L4 port of OS/2 cost such an incredible amount of money, and it produced an OS with no networking, and was dreadfully slow as well. IBM wanted BIG money to run OS/2 in SMP, meanwhile NT workstation supports two processors out of the box. You can guess which I was running on my dual proc P100.
With NT you run basically the same OS on the desk and the server, so for many dev's to make a 'server' version was all too easy. And compared to NT, OS/2 was a horrible server. I'd take NT's registry over the insane config.sys any day. Not to mention one goof in config.sys and you can't boot.
OS/2 could have been made to become more NT like, but IBM clearly wasn't up to the task, instead they were basically maintaining the same codebase from MS OS/2 2.0 circa 1991.
*YES WHY NOT*. PCtask was insanely slow, did you ever use it? And bridgeboards were not only as expensive as a PC, but an Amiga with a 386sx bridgeboard would cost far more than a decent 486 of the era.
AmigaDOS had no memory protection and no resource management. If your program crashed, and left filehandles open you had to reboot. If it overwrote anything important, off to guru land. And good lord, ever setup AmiTCP? Not exactly a walk in the park.
So what really killed the Amiga? It's simple, you could ONLY get them from Commodore. It simply couldn't compete with the Taiwanese factories cranking out XT/AT/386 boards, CGA/EGA/VGA adapters. And they sure weren't going to license Angus/Denise/Paula/Buster and friends to any 3rd parties, along with AmigaDOS. Instead Commodore was more about a pump and dump stock manipulation scheme that finally caught up with them when they were blocked from importing the CD-32 over some stupid blinking xor patent.
Also the Amiga 1000/2000/500/600 all had the same woefully underpowered 7Mhz 68000. The unexpandible models were a joke, and the 3rd party upgrade CPU modules were horribly unstable at best. (I've had several for an Amiga 2000, 3000 and a 600).
Commodore and the Amiga were always doomed to failure, even if they didn't screw up the Amiga 3000/UX deal with SUN, their reliance on trying to own the entire process, and being closed off, along with a hopelessly dated, and un-networkable OS was going to doom them anyways.
So ture. A/UX was Apple's best hope for a decent OS, but they always seemed to do their best to ignore it. It was pretty awesome back in the day, SYSV Unix, with OS 7 finder + apps. Hell even softpc ran on A/UX.
So instead of making more Mac's with full 68040's they did all these LC crap things. Then on the move to PowerPC, they went with AIX of all things on one model but had no MacOS compatibility at all. A/UX was more like OS X today, an they could have had it everywhere in the early 90's but they were more contended with their craptastic OS 7/8/9 at the time.
That was always a 'team os/2' stumbling block. The OS/2 guys belonged to one division of IBM, and the PC guys were another division. The best they would do is set it up for dualboot, but DOS/Windows was the default, I remember people with IBM PC's that went out to buy OS/2 to be surprised when they found out they already had it.
Well not only is the best way to run MS-DOS in the 1990's on OS/2 2.0, but the best way to run a Windows app in it's own copy of Windows was OS/2. IBM did themsleves no favours by charging a fortune for the SDK, and tools, nor was forcing this SAA crap on OS/2 instead of directly using Microsoft Windows on OS/2 like MS had wanted to.
There was that skunkswork project, WLO http://en.wikipedia.org/wiki/W... which of all things ended up being the basis for Win32 on NT once they dumped the OS/2 cruiser personality.
It was amazing when NT 4.0 shipped how everyone turned around and killed their Netware servers. Of course the whole TCP/IP native services thing helped, and the far cheaper licensing for NT didn't hurt either, and then there was that netware emulation package for NT that was pretty awesome too.
that being said, I've helped a handful of people migrate their old Netware 3/4 stuff onto KVM, and it's kind of funny seeing it running on 'modern' hardware.
MacOS didn't get a hypervisor until the advent of the BlueBox in 1999. But it's more a fault of the Motorola processors lacking something like v86 mode.
Windows/386 was amazing for the timeframe. Back in 1987 you could run MS-DOS boxes, *IN A WINDOW*. If you were rich enough to have EGA or VGA, you could play CGA games.. IN A WINDOW.
It was amazing, compared to other things out there. Plus there was the few dozen apps, like Word and Excel.
Windows NT was really ahead of it's time, born in 1988, and shipped in 1993. It's funny reading about it in showstopper (http://www.amazon.com/Show-Stopper-Breakneck-Generation-Microsoft/dp/0029356717) how they were surprised how deadly slow it was when they finally went to test it on actual hardware.
But CPUs got faster, memory got cheaper and here we are in a SYSv clone, Mach/BSD derivative, and NT world.
But for me, NT 3.5 was the first killer NT which was WAY faster than 3.1, and had awesome PPP support, unlike OS/2 which pretended that networking and this internet thing wasn't a thing.
That said, I have a copy of OS/2 2.0, along with the TCP/IP pack now, and it's so basic. Again no file serving capabilities, to make OS/2 even close to NT 3.1's level of usefulness. IBM really goofed by leaving networking out of the equation.
And of course the usual complaints, IBM wanting way too much money for the SDK/tools, and making 1.x too 286 centric, while not letting MS put windows on top of OS/2.
IBM shot OS/2 in the head when they announced it alongside the PS/2 in April of 1987.
Microsoft had Windows/386 in November/December 1987. Think about that, 'EU DOS 4.0' aka where OS/2 came from was still in real mode, while MS had a 386 hypervisor that they were shipping out the Compaq before the end of '87. By forcing MS to keep OS/2 on the 286 without any 386 based features, and charging $2000+ for a SDK OS/2 was dead before 1.0 was even close to GA. And releasing 1.0 without the UI was a major disaster, 1.1 should have been the first public offering. 1.0 should have been given out for free along with the SDK to developers.
But that's IBM thinking they can squeeze both ends of the toothpaste, dreaming they were the only game in town. Windows 3.0 showed Microsoft that they didn't need an IBM partnership anymore, and that their 'good enough' software was 'good enough' to sell on their own, and in their own direction.
Cheap software tools is what made Windows. While IBM was demanding $2000+ for an OS/2 SDK, MS was willing to give the SDK away to people who bundled it with their tools, and of course we had the $99 era of compilers including Visual Basic, Quick C, Turbo C and others.
OS/2 1.x did not have any cheap/discount compilers.
OS/2 had a horrible adoption rate, and the single MS-DOS penalty box was disastrous for compatibility. IBM forcing Microsoft to make it run on the 286 was a complete waste of time. In 1987 Windows/386 shipped, and it was far more useful than OS/2, simply because it truly could multitask MS-DOS, you know where the applications were.
The only thing more boneheaded by making OS/2 for the 286, was to charge well over $2000 USD for the privilege of a SDK. This lesson was of course quickly learned by MS, who would charge $99 for Windows / NT, and or give them away at any given chance. And then there was the hobbiest market with products like Quick C for Windows, and even Borland's Turbo C++ for Windows. There was nothing like this for OS/2 where development tools were big $$, until it went 32bit and EMX entered the scene with GCC.
this is where Microsoft broke away from being an IBM partner, to take control of their own destiny. IBM had effectively killed OS/2 with it's insane SDK prices, and per seat costs. Not to mention the complete lack of applettes, and by refusing to let Microsoft do anything with the UI, or allow for OS/2 to run windows binaries. But the success of Windows 3.0 changed all of that.
What did Windows 3.0 give us? Well, while Windows/386 was a really cool 386 hypervisor, Windows itself, and all windows programs were restricted to 640kb of real mode memory. But Windows 3.0 was built around a MS-DOS extender, and now you could run in protected mode on a 286/386. And even better you didn't have to change your OS, just install Windows and go. Not to mention since it sat on DOS, you could still use MS-DOS based drivers, TSR's. It was simply a massive thing. Also licensing MS-DOS extenders at the time was VERY expensive, and per application. Writing a Windows application, along with the license costs of Windows 3.0 was much cheaper.
From this point MS's OS/2 3.0 project became Windows NT, and MS pulled away from the deathmarch project that was OS/2 2.0. The funny thing is that OS/2 2.0 was delayed to add in the most confusing shell (to users, I know programmer's and tech people loved WPS, but to average users, it was a nightmare) and Windows compatibility via specialized drivers. Things that MS wanted to do, but IBM refused to let them.
The sad thing is that bringing Windows up to some kind of usable level where OS/2 was basically already, and by making 286 processor based machines useful ended up setting us back a good 5+ years until the Windows 95 avalanche finally pushed 32bit computing to the masses. Although it wasn't until 2001 with XP Home did it finally become truly usable.
NT, while being a solid future looking design was at the time so massive, and so complex that running it on a 386 was a horrible experience. But as processors got faster the NT investment eventually paid off, with NT being found almost everywhere these days.
So yes, Windows 3.0 was the most significant product Microsoft shipped, that ended up not only defining the direction of the company, but also the industry. Finally everyone could unlock the power of their 286+ computer that was basically un-used by MS-DOS.
Slashdot Mirror
Are you still on VAXen? I thought the Itanium stuff was at least newer than the Dec Alpha...
http://www.cs.utah.edu/flux/os...
so cool back in the day, but it didn't really take hold for some reason. too bad.
at least the name superfish isnt missleading, if not ironic.
give us the butterfly! The only cool IBM laptop.
Blue enter key? Are you kidding me?
Also, some OS/2
check out shoebill ( http://emaculation.com/forum/v... ), this guy managed to get A/UX run under emulation. It's pretty cool
SAA is what OS/2's presentation manager was built around. Some pipe dream that the mainframes, as/400's and rs/6000's were going to share a common UI. Well that never happened, and it was a kneejerk thing to push MS out of the UI on OS/2.
Not that it matters, MS-DOS had a lot more device drivers than OS/2, and Windows 3.0's ability to use them made it a winner.
OS/2 was more of a learning tool in how not to push people off of MS-DOS, and instead they moved to Windows, then once machines were fast enough and ME was horrible enough, everyone went to XP Home, and plenty of users are still there.
They kind of did with LanMan server, and things like SQL. The OS/2 Extended Edition bundled lots of stuff together, but it was IBM's way of doing things, and I never saw anyone using EE. However I've setup MS SQL 1.0 on OS/2 and it is a NIGHTMARE. Compared to NT where you install NT, then SQL and away you go. But no, Install OS/2, reboot install the lan driver stuff, reboot, install lan man server, reboot make sure you can now create named pipes, and read them, then install sql server. OS/2 refused to bundle in the important bits, that NT and WfW later would all bring in by default. It didn't take a genius to see the rise of the LAN, however it took some major pushes to get into server space. But nobody enjoyed dealing with netware and their NLM crap, once NT hit v 4.0 everyone was dumping it for NT. But the underpinning network aspects of NT were on OS/2 + LanMan. NT just started from that point and had it all built in.
What is weird is how MS saw the LAN, but missed the internet. Even OS/2 3.0 Warp with that woefully useless IAK, provided no LAN access, or any peer to peer networking capabilities.
Granted the window between 3.0 and 3.1 for workgroups was a crappy period for networking but by the time OS/2 3.0 came out, it was already too late. And it's TCP/IP was that horrible dialup centric POS, nothing multi-homed, or with physical network cards.
in the OS/2 1.x there was a clear leader in the LAN, and it was Netware. But IBM just thought OS/2 should be as crappy as MS-DOS, and let the 3rd parties come along and add in support. Instead they should have at least bundled in LanMan support, but of course that would have killed a part number.
Windows for Workgroups 3.1 did have a network stack, and it was greatly improved in Windows for Workgroups 3.11. NT 3.1 came with TCP/IP, IPX, and support for SNA out of the box in 93. Warp was in '94.
The real 'killer' of course was Win32s+Winsock which gave us Mosaic. And why was IBM not killing themselves to make a Mosaic port to OS/2? Once people saw a graphical Internet, everyone I saw around me was clamouring for it, and it was amazing at the time how many Windows 3.1+Win32s+Mosiac installs talking to various UNIX dialup accounts using SLiRP. Something that was really unstable on OS/2, so more of the OS/2 fans were either going to Linux or to NT 3.5/3.51
It wouldn't matter, as SMP was becoming a thing, and don't forget the coming x86_64 along with the ability to run on RISC. OS/2's kernel was largely untouched from early MS OS/2 2.0 betas, and the device drivers were still 16bit assembly. IBM's L4 port of OS/2 cost such an incredible amount of money, and it produced an OS with no networking, and was dreadfully slow as well. IBM wanted BIG money to run OS/2 in SMP, meanwhile NT workstation supports two processors out of the box. You can guess which I was running on my dual proc P100.
With NT you run basically the same OS on the desk and the server, so for many dev's to make a 'server' version was all too easy. And compared to NT, OS/2 was a horrible server. I'd take NT's registry over the insane config.sys any day. Not to mention one goof in config.sys and you can't boot.
OS/2 could have been made to become more NT like, but IBM clearly wasn't up to the task, instead they were basically maintaining the same codebase from MS OS/2 2.0 circa 1991.
*YES WHY NOT*. PCtask was insanely slow, did you ever use it? And bridgeboards were not only as expensive as a PC, but an Amiga with a 386sx bridgeboard would cost far more than a decent 486 of the era.
AmigaDOS had no memory protection and no resource management. If your program crashed, and left filehandles open you had to reboot. If it overwrote anything important, off to guru land. And good lord, ever setup AmiTCP? Not exactly a walk in the park.
So what really killed the Amiga? It's simple, you could ONLY get them from Commodore. It simply couldn't compete with the Taiwanese factories cranking out XT/AT/386 boards, CGA/EGA/VGA adapters. And they sure weren't going to license Angus/Denise/Paula/Buster and friends to any 3rd parties, along with AmigaDOS. Instead Commodore was more about a pump and dump stock manipulation scheme that finally caught up with them when they were blocked from importing the CD-32 over some stupid blinking xor patent.
Also the Amiga 1000/2000/500/600 all had the same woefully underpowered 7Mhz 68000. The unexpandible models were a joke, and the 3rd party upgrade CPU modules were horribly unstable at best. (I've had several for an Amiga 2000, 3000 and a 600).
Commodore and the Amiga were always doomed to failure, even if they didn't screw up the Amiga 3000/UX deal with SUN, their reliance on trying to own the entire process, and being closed off, along with a hopelessly dated, and un-networkable OS was going to doom them anyways.
So ture. A/UX was Apple's best hope for a decent OS, but they always seemed to do their best to ignore it. It was pretty awesome back in the day, SYSV Unix, with OS 7 finder + apps. Hell even softpc ran on A/UX.
So instead of making more Mac's with full 68040's they did all these LC crap things. Then on the move to PowerPC, they went with AIX of all things on one model but had no MacOS compatibility at all. A/UX was more like OS X today, an they could have had it everywhere in the early 90's but they were more contended with their craptastic OS 7/8/9 at the time.
That was always a 'team os/2' stumbling block. The OS/2 guys belonged to one division of IBM, and the PC guys were another division. The best they would do is set it up for dualboot, but DOS/Windows was the default, I remember people with IBM PC's that went out to buy OS/2 to be surprised when they found out they already had it.
Well not only is the best way to run MS-DOS in the 1990's on OS/2 2.0, but the best way to run a Windows app in it's own copy of Windows was OS/2. IBM did themsleves no favours by charging a fortune for the SDK, and tools, nor was forcing this SAA crap on OS/2 instead of directly using Microsoft Windows on OS/2 like MS had wanted to.
There was that skunkswork project, WLO http://en.wikipedia.org/wiki/W... which of all things ended up being the basis for Win32 on NT once they dumped the OS/2 cruiser personality.
It was amazing when NT 4.0 shipped how everyone turned around and killed their Netware servers. Of course the whole TCP/IP native services thing helped, and the far cheaper licensing for NT didn't hurt either, and then there was that netware emulation package for NT that was pretty awesome too.
that being said, I've helped a handful of people migrate their old Netware 3/4 stuff onto KVM, and it's kind of funny seeing it running on 'modern' hardware.
Informal English, do you type it?
MacOS didn't get a hypervisor until the advent of the BlueBox in 1999. But it's more a fault of the Motorola processors lacking something like v86 mode.
Windows/386 was amazing for the timeframe. Back in 1987 you could run MS-DOS boxes, *IN A WINDOW*. If you were rich enough to have EGA or VGA, you could play CGA games.. IN A WINDOW.
It was amazing, compared to other things out there. Plus there was the few dozen apps, like Word and Excel.
Windows NT was really ahead of it's time, born in 1988, and shipped in 1993. It's funny reading about it in showstopper (http://www.amazon.com/Show-Stopper-Breakneck-Generation-Microsoft/dp/0029356717) how they were surprised how deadly slow it was when they finally went to test it on actual hardware.
But CPUs got faster, memory got cheaper and here we are in a SYSv clone, Mach/BSD derivative, and NT world.
But for me, NT 3.5 was the first killer NT which was WAY faster than 3.1, and had awesome PPP support, unlike OS/2 which pretended that networking and this internet thing wasn't a thing.
That said, I have a copy of OS/2 2.0, along with the TCP/IP pack now, and it's so basic. Again no file serving capabilities, to make OS/2 even close to NT 3.1's level of usefulness. IBM really goofed by leaving networking out of the equation.
And of course the usual complaints, IBM wanting way too much money for the SDK/tools, and making 1.x too 286 centric, while not letting MS put windows on top of OS/2.
Desqview had no applications. by the time they bothered with a SDK, windows 3.0 was already a thing.
They probably didn't have the fortitude to sit though through the training video.
https://youtu.be/aoBEYUFagi8
IBM shot OS/2 in the head when they announced it alongside the PS/2 in April of 1987.
Microsoft had Windows/386 in November/December 1987. Think about that, 'EU DOS 4.0' aka where OS/2 came from was still in real mode, while MS had a 386 hypervisor that they were shipping out the Compaq before the end of '87. By forcing MS to keep OS/2 on the 286 without any 386 based features, and charging $2000+ for a SDK OS/2 was dead before 1.0 was even close to GA. And releasing 1.0 without the UI was a major disaster, 1.1 should have been the first public offering. 1.0 should have been given out for free along with the SDK to developers.
But that's IBM thinking they can squeeze both ends of the toothpaste, dreaming they were the only game in town. Windows 3.0 showed Microsoft that they didn't need an IBM partnership anymore, and that their 'good enough' software was 'good enough' to sell on their own, and in their own direction.
Cheap software tools is what made Windows. While IBM was demanding $2000+ for an OS/2 SDK, MS was willing to give the SDK away to people who bundled it with their tools, and of course we had the $99 era of compilers including Visual Basic, Quick C, Turbo C and others.
OS/2 1.x did not have any cheap/discount compilers.
OS/2 had a horrible adoption rate, and the single MS-DOS penalty box was disastrous for compatibility. IBM forcing Microsoft to make it run on the 286 was a complete waste of time. In 1987 Windows/386 shipped, and it was far more useful than OS/2, simply because it truly could multitask MS-DOS, you know where the applications were.
The only thing more boneheaded by making OS/2 for the 286, was to charge well over $2000 USD for the privilege of a SDK. This lesson was of course quickly learned by MS, who would charge $99 for Windows / NT, and or give them away at any given chance. And then there was the hobbiest market with products like Quick C for Windows, and even Borland's Turbo C++ for Windows. There was nothing like this for OS/2 where development tools were big $$, until it went 32bit and EMX entered the scene with GCC.
this is where Microsoft broke away from being an IBM partner, to take control of their own destiny. IBM had effectively killed OS/2 with it's insane SDK prices, and per seat costs. Not to mention the complete lack of applettes, and by refusing to let Microsoft do anything with the UI, or allow for OS/2 to run windows binaries. But the success of Windows 3.0 changed all of that.
What did Windows 3.0 give us? Well, while Windows/386 was a really cool 386 hypervisor, Windows itself, and all windows programs were restricted to 640kb of real mode memory. But Windows 3.0 was built around a MS-DOS extender, and now you could run in protected mode on a 286/386. And even better you didn't have to change your OS, just install Windows and go. Not to mention since it sat on DOS, you could still use MS-DOS based drivers, TSR's. It was simply a massive thing. Also licensing MS-DOS extenders at the time was VERY expensive, and per application. Writing a Windows application, along with the license costs of Windows 3.0 was much cheaper.
From this point MS's OS/2 3.0 project became Windows NT, and MS pulled away from the deathmarch project that was OS/2 2.0. The funny thing is that OS/2 2.0 was delayed to add in the most confusing shell (to users, I know programmer's and tech people loved WPS, but to average users, it was a nightmare) and Windows compatibility via specialized drivers. Things that MS wanted to do, but IBM refused to let them.
The sad thing is that bringing Windows up to some kind of usable level where OS/2 was basically already, and by making 286 processor based machines useful ended up setting us back a good 5+ years until the Windows 95 avalanche finally pushed 32bit computing to the masses. Although it wasn't until 2001 with XP Home did it finally become truly usable.
NT, while being a solid future looking design was at the time so massive, and so complex that running it on a 386 was a horrible experience. But as processors got faster the NT investment eventually paid off, with NT being found almost everywhere these days.
So yes, Windows 3.0 was the most significant product Microsoft shipped, that ended up not only defining the direction of the company, but also the industry. Finally everyone could unlock the power of their 286+ computer that was basically un-used by MS-DOS.
Is it made out of bakelite? I hope it has a dial, none of this DTMF crap either!