15 thoughts on “Yuhong Bao’s post on the OS/2 2.0 fiasco”
Uhm.. Yuhong is always going on about those ‘unethical attacks’ and ‘the Munchins’ and all that, in posts all over the net. And very little to back it up, it’s always just the same, over and over again. As far as I can see there weren’t anything unethical in MS leaving OS/2, IBM brought it onto themselves and MS had no possible choice in the matter even if MS had been the most ethical company ever seen (and that MS presumably isn’t simply doesn’t matter one way or another in this case).
It’s enough to refer to earlier references here on virtuallyfun to see that, e.g. the register articles from IBM insiders, the old Usenet post from an MS insider.
The “Munchkins” goes back to John C Dvorak naming it..
Although after all the dust settled years later, I find IBM more to blame about OS/2 falling through, it is pretty clear MS wanted to do a 386 version ignoring the 286, Windows/386 was a capable hypervisor back in 1987. Even back in 1987 there was a 386 version of Microsoft C (although targeted for Xenix), the PC could have been enjoying something more .. NT like back in the late 1980’s with an OS/2 base, and Windows API much like what NT became..
And basically IBM rode OS/2 2.0 into the ground with a few “improvements” here and there, but OS/2 held onto config.sys, being hybrid 16bit/32bit and insane billing modes…
And as someone mentioned in a comment on my blog posts, 386s were somewhat more expensive even back in 1990, let along in 1987 which was not long after it was first released.
Again the 286 was a major mistake, clearly MS had been working on 386 specific stuff to launch Xenix and Windows/386 in 1987.
IBM holding MS to the IBM AT was a catastrophic mistake, that set the industry back from 1987 until 1995.
Windows 3.0 & OS/2 1.x should have never happened, but IBM muscled in with insisting 286 support, and by pushing Windows out of the Windowing stack of OS/2 and insisting on this SAA nonsense that went nowhere as a major mistake.
The bottom line is that IBM defines no standards today, while Microsoft is still a player.
Well, how many ATs and clones were already sold? And to be honest on Intel’s side, the 286 and 186 was released only months after the IBM PC came out. What was a major mistake IMO is not releasing the 386SX until 1988.
Indeed, I don’t think Intel had any sense of just how important the IBM PC market was.. there was some other risc like chip intel did in between the 8086 & the 80286 that was going to change the world that went nowhere….
Instead the 80386 basically revolutionized the world by bringing 32bit computing to the masses along with hardware virtualization, while the 286 was such a … brain dead chip with so much potential, had it had a flat protected mode capable of addressing 16MB of ram, or 1GB as a 1GB segment instead of 64KB segments the 286 would have been a practical alternative.. but it took a chip like the 80386 or the 68020 which people could move code from 31bit mainframes & 32bit minis, along with a MMU to the micro level for the world to take off..
Oh and regarding initial 386 sales, don’t forget about the 16bit-swonly scandal, or the first yield running at 12Mhz.. For the first year the 386 looked like a DUD, but once Compaq signed up for it, and being a solid enough of a vendor did the market solidify for it.. In 1987 a decent 386 cost a fortune, but the benefits of targeting the 386 really were greater than any risk mitigation by even dealing with the 286 as the 386 and the 32bit architecture was clearly the direction of the future, while the 286 was some 16bit stopgap answer to the 68000 but designed in terms of a “powerful 8080” instead of a capable 16bit processor. Then again the internal registers of the 68000 were 32bit… While, of course the 80186/80286 were still 16bit.
“IBM holding MS to the IBM AT was a catastrophic mistake, that set the industry back from 1987 until 1995.”
Until 1992 at most. It was the OS/2 2.0 fiasco that set it back to 1995. For comparison, the Pentium was released in 1993.
yeah I remember it all too well, how many people were buying sooped up Pentium’s and running 16bit software, the Pentium PRO brought that to a head even in 1996, 1997 …
But we should have been there in 1987 1988 …
This also reminds me of the CMD640/RZ1000 problems where the two IDE channels provided by these controllers was not really independent. Suffice to say that if OS/2 2.x replaced DOS/Windows instead of MS turning it into an entire fiasco I am sure these chips would never have made it out the door with these problems.
Uhm.. Yuhong is always going on about those ‘unethical attacks’ and ‘the Munchins’ and all that, in posts all over the net. And very little to back it up, it’s always just the same, over and over again. As far as I can see there weren’t anything unethical in MS leaving OS/2, IBM brought it onto themselves and MS had no possible choice in the matter even if MS had been the most ethical company ever seen (and that MS presumably isn’t simply doesn’t matter one way or another in this case).
It’s enough to refer to earlier references here on virtuallyfun to see that, e.g. the register articles from IBM insiders, the old Usenet post from an MS insider.
-Tor
The “Munchkins” goes back to John C Dvorak naming it..
Although after all the dust settled years later, I find IBM more to blame about OS/2 falling through, it is pretty clear MS wanted to do a 386 version ignoring the 286, Windows/386 was a capable hypervisor back in 1987. Even back in 1987 there was a 386 version of Microsoft C (although targeted for Xenix), the PC could have been enjoying something more .. NT like back in the late 1980’s with an OS/2 base, and Windows API much like what NT became..
And basically IBM rode OS/2 2.0 into the ground with a few “improvements” here and there, but OS/2 held onto config.sys, being hybrid 16bit/32bit and insane billing modes…
Not that it excuses the MS attacks (and the “Munchkins” was not the only one).
Or change the fact that they were even worse.
And as someone mentioned in a comment on my blog posts, 386s were somewhat more expensive even back in 1990, let along in 1987 which was not long after it was first released.
targeting the 286 was just as big as a mistake as the 186. IBM shouldn’t have held anyone’s feet to the 286 fire.
What do you mean? I mean, even Compaq launched their 386 only months earlier.
Again the 286 was a major mistake, clearly MS had been working on 386 specific stuff to launch Xenix and Windows/386 in 1987.
IBM holding MS to the IBM AT was a catastrophic mistake, that set the industry back from 1987 until 1995.
Windows 3.0 & OS/2 1.x should have never happened, but IBM muscled in with insisting 286 support, and by pushing Windows out of the Windowing stack of OS/2 and insisting on this SAA nonsense that went nowhere as a major mistake.
The bottom line is that IBM defines no standards today, while Microsoft is still a player.
Well, how many ATs and clones were already sold? And to be honest on Intel’s side, the 286 and 186 was released only months after the IBM PC came out. What was a major mistake IMO is not releasing the 386SX until 1988.
Indeed, I don’t think Intel had any sense of just how important the IBM PC market was.. there was some other risc like chip intel did in between the 8086 & the 80286 that was going to change the world that went nowhere….
Instead the 80386 basically revolutionized the world by bringing 32bit computing to the masses along with hardware virtualization, while the 286 was such a … brain dead chip with so much potential, had it had a flat protected mode capable of addressing 16MB of ram, or 1GB as a 1GB segment instead of 64KB segments the 286 would have been a practical alternative.. but it took a chip like the 80386 or the 68020 which people could move code from 31bit mainframes & 32bit minis, along with a MMU to the micro level for the world to take off..
But as they say the past is.. passe.
I went through a similar discussion already with MichalN and here is a quote from a private email:
“This tells a different story: http://books.google.com/books?id=DD8EAAAAMBAJ&lpg=PP1&pg=PT20#v=onepage&q&f=false
According to that report, the sales of 386s in ’87 were pretty negligible. The percentage of the total installed base must have been quite low.
The article about 386SX systems here http://books.google.com/books?id=ezAEAAAAMBAJ&lpg=PT30&pg=PP1 indicates that only at the end of ’89 did 386 sales overtake 286s. And again that’s the sales of new machines, not the total installed base.”
Oh and regarding initial 386 sales, don’t forget about the 16bit-swonly scandal, or the first yield running at 12Mhz.. For the first year the 386 looked like a DUD, but once Compaq signed up for it, and being a solid enough of a vendor did the market solidify for it.. In 1987 a decent 386 cost a fortune, but the benefits of targeting the 386 really were greater than any risk mitigation by even dealing with the 286 as the 386 and the 32bit architecture was clearly the direction of the future, while the 286 was some 16bit stopgap answer to the 68000 but designed in terms of a “powerful 8080” instead of a capable 16bit processor. Then again the internal registers of the 68000 were 32bit… While, of course the 80186/80286 were still 16bit.
“IBM holding MS to the IBM AT was a catastrophic mistake, that set the industry back from 1987 until 1995.”
Until 1992 at most. It was the OS/2 2.0 fiasco that set it back to 1995. For comparison, the Pentium was released in 1993.
yeah I remember it all too well, how many people were buying sooped up Pentium’s and running 16bit software, the Pentium PRO brought that to a head even in 1996, 1997 …
But we should have been there in 1987 1988 …
This also reminds me of the CMD640/RZ1000 problems where the two IDE channels provided by these controllers was not really independent. Suffice to say that if OS/2 2.x replaced DOS/Windows instead of MS turning it into an entire fiasco I am sure these chips would never have made it out the door with these problems.