Windows 3.1 Build list 2.0

[Windows Alfie]

Hello, everybody, I have something to present to you people.

Andy may place these screenshots into the Photo Gallery, but I do not intead for these to be in the PG/Wiki.

Here is the list. (Leaked) (TheCollectionBook)

Build 26 (5th April 1991)
Build 34e (7th June 1991)
Build 43e (27th August 1991)
Build 61d (17th December 1991)
Build 68 (3rd February 1992)
RTM (6th April 1992)

Proven to Exist, not leaked. (Wiki)

Build 14 (Proven in Antitrust Document PX05218) (Most likely January 1991)
Build 31 (Proven in Antitrust Document PX03461) (July 1991)
Build 32w (Proven by build number in Windows for Workgroups 3.1 RTM)
Build 33 (Proven by SETUP31.VER in Early Chicago and MSDOS.EXE in Build 34e.)
Build 34f (Proven by Serbian magazine) (July 1991)
Build 38 (Proven by Antitrust Document PX08861) (July 1991)
Build 40 (Proven by build number in Build 43e to Early Chicago builds in OLDSETUP.INF)
Build 43c (Proven in MS-DOS executive in Build 43e)
Build 43d (Proven by Screenshot and Antitrust Document PX05353) (August 1991)
Build 47 (Proven by Antitrust Document PX03475) (Most likely September 1991)
Build 50 (Proven by Antitrust Document PX07624) (Most likely September 1991)
Build 55 (Proven by Antitrust Document PX01146) (Most likely December 1991)
Build 58 (Proven by Antitrust Document PX01168) (November 1991)
Build 60 (Proven by Antitrust Document PX01168) (December 1991)
Build 61 (Proven by Antitrust Documents PX01168 and PX01251) (Most likely January 1992)
Build 61b (Proven by Antitrust Document PX01146) (Most likely December 1991)
Build 69 (Proven by Antitrust Document PX09409) (Most likely January 1992)
Build 70 (Proven by French Microsoft KB)
Build 71 (Proven by French Microsoft KB and PX01251) (Most likely February 1992)
Build 101 (Proven by OLECLI.DLL build in WFW3.11 411 and Antitrust Document PX01251)
RC1 (Proven by Program Review) (February 1992)
RC2 (Proven by Antitrust Document PX01251) (Most likely February 1992)
RC3a (Proven by Antitrust Document PX01251) (Most likely February 1992)
RC3b (Proven by Antitrust Document PX01251) (Most likely February 1992)
RC3c (Proven by Antitrust Document PX01251) (Most likely February 1992)
RC3d (Proven by Antitrust Document PX01251) (Most likely February 1992)
RC3e (Proven by Antitrust Document PX01251) (Most likely February 1992)
RC3f (Proven by Antitrust Document PX01251) (Most likely February 1992)
RC3g (Proven by Antitrust Document PX01251) (Most likely February 1992)


Will be updated soon to include more screenshots.

-Windows Alfie

[DiskingRound]

Add the unleaked builds 34f, which is referenced in some computer magazine according to the wiki, and 70, which is referenced in some MS article according to the wiki.

[Battler]

- Windows Alfie: Are you sure your copy of 068 isn't the one that's floating around the Internet that is frankenbuild of 061d, RTM, and some 068 files?

[Windows Alfie]

- Battler:

Not too sure, found it at WinWorld.

If it is real, let me know.


- DiskingRound:

Just been updated my friend!

[os2fan2]

Is Win-OS/2 3.1 in the list?

[Windows Alfie]

- os2fan2:

OS2 Is not counted here, since this is 3.1, not NT 3.1.

[The Distractor]

- os2fan2:

OS2 Is not counted here, since this is 3.1, not NT 3.1.

No, he's referring to Win-OS/2 3.1, which is the version of Windows 3.1 bundled with some versions of OS/2 2.1 and above.

[Windows Alfie]

- The Distractor:

What I mean is I only do the main builds, not any sub builds like: Fakes, WfW3.1 and 3.11.

I only list the main builds that started other OS types running on the 3.1 version.

[johnleakedfan]

Thanks for making this list, I bet this will continue to grow. Usually it is alot harder to find new info on unleaked builds of pretty old OS's (or new betas in general). Only if someone uploads a legit copy of an unleaked build, of course

[Windows Alfie]

- johnleakedfan:

I made for all them 3.1 historians as a brief timeline, hoping in the sense that would be a nice feature to BetaArchive.

This will be uploaded daily.

Builds that are unknown will be put in the unknown section if leaked.

If it is a true leak, it will go into the Leaked category.

If a false leak, It will not be mentioned here.

Also, thanks for positive feedback!

[Battler]

Build 32w (Proven by build number in Windows for Workgroups 3.1 RTM)

No, that's just the version of the SETUP.INF file format.

Build 40 (Proven by build number in Build 43e to Early Chicago builds in OLDSETUP.INF)

That's, again, just the version of the SETUP.INF file format.

Build 33 (Proven by SETUP31.VER in Early Chicago and MSDOS.EXE in Build 34e.)

That's 3.10.2.033, but also identifies itself as "Windows version 3.20 (Cougar)". This means it was used in the Cougar project, but was forked from Windows for Workgroups 3.1 build 033.

[WindowsCollector2000]

That's 3.10.2.033, but also identifies itself as "Windows version 3.20 (Cougar)". This means it was used in the Cougar project, but was forked from Windows for Workgroups 3.1 build 033.

Offtopic Comment

And to think that Cougar was some fake floating around in the Internet...

[Overdoze]

Offtopic Comment

And to think that Cougar was some fake floating around in the Internet...

It was, actually. Some fake thing called Cougar build 28 was widely spread years ago and can still be found today.

[Battler]

The leaked Cougar was a fake obviously, however real Cougar existed, but it was more like a 32-bit protected mode DOS rather than Windows. The remnants of it in Chicago are that WinVer thing, and the fact the 32-bit kernel is called DOS386.EXE.

[Windows Alfie]

BTW, some people confused me here since the info is from the Wiki. I just list what's in the Wiki in brief.

[Overdoze]

Not sure what's confusing here... The Cougar project is real but was never intended to be an actual standalone release of Windows, together with Jaguar it was part of Chicago. Like Battler said, that 3.10.2.033 or whatever string is a remnant of that and not a Windows 3.1 build - Wiki tends to have outdated info sometimes.

That said, Cougar build 28 that you can find online is a fake. Aside from being a modded Windows 3.1 IIRC, it also completely missed the point of the actual Cougar (not a Windows release).

[Battler]

The Cougar project is real but was never intended to be an actual standalone release of Windows, together with Jaguar it was part of Chicago.

Actually, wasn't it supposed to be a standalone 32-bit DOS before being integrated into Chicago? At least that's what I remember the Anti-Trust documents saying.

[Overdoze]

Don't quote me on that, it's been quite a while since I last read the AT docs, would need to check again to be sure...

[badmoon]

Actually, wasn't it supposed to be a standalone 32-bit DOS before being integrated into Chicago?

According to Raymond Chen's blog, Cougar was indeed meant to be a 32-bit DOS kernel; and its code was later used to develop VMM32, an important component of Win9x OSes.

However, Cougar wasn't meant to be a standalone 32-bit DOS, as such. The idea was that Jaguar—a modified 16-bit DOS—could use the 32-bit Cougar. They would make up the new DOS environment together.

[jimmsta]

I had always thought that Cougar was effectively the combination of IO.SYS and the VMM32 subsystem. That is to say, the DOS 16-bit kernel stored in IO.SYS + the 32-bit virtual machine that executes services on the win32 implementation in the 9x kernel (effectively Jaguar).

[xelloss]

I had always thought that Cougar was effectively the combination of IO.SYS and the VMM32 subsystem. That is to say, the DOS 16-bit kernel stored in IO.SYS + the 32-bit virtual machine that executes services on the win32 implementation in the 9x kernel (effectively Jaguar).

Try replacing KRNL386.EXE with COMMAND.COM: that should be a good approximation of Cougar.

[Hyoenmadan86]

I had always thought that Cougar was effectively the combination of IO.SYS and the VMM32 subsystem. That is to say, the DOS 16-bit kernel stored in IO.SYS + the 32-bit virtual machine that executes services on the win32 implementation in the 9x kernel (effectively Jaguar).

Except that VMM is effectively a full kernel (hypervisor type i would say, term which didn't exist in these days) running a DOS environment as compatibility layer (and V86 mode), as explained in literature covering internals of windows versions based in the VMM. Here DOS16 kernel task is work as bootloader, and the compatibility layer, as described by the MS engineer in the article.

So nope, VMM isn't just a "Virtual Machine executing services". Also many examples of such hybrid technologies exist besides Windows386 based OSs. Things like Netware for OS/2 and IGC VM/386, ancient technologies which in last instance evolved to actual stuff like ESXi or Xen.

[jimmsta]

I had always thought that Cougar was effectively the combination of IO.SYS and the VMM32 subsystem. That is to say, the DOS 16-bit kernel stored in IO.SYS + the 32-bit virtual machine that executes services on the win32 implementation in the 9x kernel (effectively Jaguar).

Try replacing KRNL386.EXE with COMMAND.COM: that should be a good approximation of Cougar.

Well, that's pretty neat.

Code: Select all

-----Memory map------------------------------------------------
020A-05DE  03D50 bytes  Owner=0008
05E0-05E3  00040 bytes  Owner=0008
05E5-05EF  000B0 bytes  WIN
05F1-06C5  00D50 bytes  WIN
06C7-06C8  00020 bytes  VMM32
06CA-083B  01720 bytes  VMM32
083D-083F  00030 bytes  KRNL386
0841-09A5  01650 bytes  KRNL386
09A7-09FF  00590 bytes  KRNL386

Size of DOS.....................................8K bytes
Amount of allocated memory......................31K bytes
Amount of free memory...........................596K bytes
Total amount of memory..........................635K bytes

Generated this using the pharlap utility tellme, included in early versions of Microsoft Visual C. This is running within the replaced krnl386.exe/command.com shell.