A bunch of the SDK is going on about how obsolete so many of the old calls were as you don’t have to mess with segmentation and all that jazz anymore. Sometimes I wonder with the WLO effort, if there was some more thought of bypassing PM all together and porting Windows directly to OS/2. According to the Letwin email, all this Windows on OS/2 really pissed IBM off to the point of IBM wanting MS to give up on Windows… Which if you look at the MS OS/2 applications Excel, Word for PM & Mail were all using WLO to some degree. Like their app people didn’t want to write for Windows, but rather wanted to ‘port’ Windows apps…

No doubt a lot of lessons were learned in moving Win16 to Win32… In more that Win32 is ‘similar’ to Win16, but isn’t.

The real shame is that I’m pretty sure that OS/2 2.0 is more like Windows 3.0 + Win386 (the Watcom Windows extender) than like a full 32bit OS. There is some article that mentions that the first 32bit version is just something that extended OS/2 1.2 …

I guess it also goes onto why there is no OS/2 freeware clones.. like why bother? the ‘best’ stuff is what? MS 16bit stuff? I’m sure the Fortran junkies moved on to PowerStation and beyond. 16bit protected mode Basic, or Pascal? The 32bit OS/2 stuff was few and far between.

From what I understand there was a MSC 5.0,5.1 and a 5.2. The 386 version of MSC dates back to Xenix on the 386.

from the 386 read.me

If an application is to remain 16-bit, it should be
compiled and linked with the 1.2 toolkit header files
and libraries. If it is compiled with a 16-bit
compiler (c5.1/c5.2/c6.0) and OS/2 2.0 SDK
header files, and linked with the 2.0 SDK’s libraries,
it is probable that the application will have
references to API’s that are not in the same system DLL
in the 2.0 system that they were in the 1.2 system.
This will cause the 1.2 system to refuse to load the
application because it is unable to resolve the dynamic
links.

Which means that OS/2 1.2 and the 2.0 beta were not yet in sync. Also from what I remember in showstopper, Cutler and crew were hired on at the end of 1988, in 1989 they were designing NT OS/2, and in 1990 they had kernel mode stuff working and in the middle of the year they dumped the cruiser ‘OS/2 32bit’ personality and took the port of Windows 3.0 to OS/2 (WLO) as the starting point for Win32 on NT. I don’t think NT OS/2 was to the point of anything to show off, as MS would have done so at this point. But they were showing off Windows NT at the close of 1991 (At the October comdex, the same time IBM was giving out the OS/2 2.0 LA).

And as we all know, OS/2 2.0 shipped in 1992, and NT in 1993.

While the version number is higher than 1.00.075, the build number is lower, the copyright years match (1989). My guess is, that at some point, when OS/2 and NT were developed concurrently, there were two branches of the 386 C compiler.

*cough* bbs… or vpsland…

Do you plan on uploading said c compiler + SDK somewhere?