Well, as for the October 1991 release and Build 239.1 of Windows NT 3.2 using the same collection of wallpaper from Windows 3.00a, well, the reason for this is that much of the graphical user interface is the same as that of Janus (Windows 3.1) Build 34e from June 7th, 1991 (Build 34e also used the same wallpaper collection as Windows 3.00a), also, the Windows NT 3.2 builds are behind in features, for example, in the case of the October 1991 release and Build 239.1, the graphical user interface is a cross between that of Windows 3.00a, Janus Build 34e, and Build 61d, even though Janus Build 61d, having been released in December 1991, had already progressed significantly in terms of its graphical user interface and general features.

The most likely explanation for the Windows NT 3.2 builds being behind in features is the time already spent in writing, compiling, and testing the low level kernel, which is quite different from the real mode (KERNEL.EXE), DOS Extender (DOSX.EXE), and VMM (WIN386.EXE) low level kernels in the Windows 3.x builds, not to mention the time spent in adapting the actual feature set normally used in the Windows 3.x builds to the Windows NT 3.2 operating system, which certainly explains why the earlier builds of Windows NT that used the version number 3.2 (and as far as I know, also 3.0 earlier, though it could have also been a mistyping) are behind in features.

Even as recently as Build 319.1 (October 1992), when the version number had already been changed from 3.2 to 3.1, Windows NT 3.1 still had some aspects of Windows 3.00a's color scheme, not to mention that even as recently as Build 404 and as far as I know, even Build 438 (March 1993), the Control Panel icon was still that of the earlier Janus (Windows 3.1) builds, and not that of the final Windows 3.1, but that it was updated to match that of the final Windows 3.1 by the time Windows NT 3.1 was released.

Similarly, the earlier Cairo (post-RTM Windows NT 3.5, later, the version number shifted to 4.0) builds were also behind in features in terms of the Explorer shell, and even by September 1995 with the recent release of Windows 95 and the compiling of Build 1130, the Explorer shell and updated graphical user interface still retained some aspects of Chicago (by then already named Windows 95) Build 189's user interface and shell, although they too, were updated by the 12xx series builds of which were released in January 1996.

You may want to look again....



Sure it's not terribly intuitive to use, like all these old things, but the C compiler is certainly there. At the same time for 10 megabytes it sure packs a lot of stuff into it.

I xcopied it onto HPFS.. probably more 'dangerous' but certainly more fun.

Yeah, file command confirms it.

# file NTOSKRNL.EXE
NTOSKRNL.EXE: 80386 COFF executable not stripped - version 1



It won't run because the PE header differs.

# file CALC.EXE
CALC.EXE: PE for MS Windows (unknown subsystem) unknown processor

subsystem and processor in PE header are unknown, as a result Windows refuse to recognize it as PE executable and run it in NTVDM.

I like these old stuff!

I donno feels too much like the WRK...

Microsoft Windows(R) NT WRK Community Kernel [v6.0.5001]
Init: System initializing...
(..\initos.c:1430) SystemMemory: 130536Kb, 1 System Processor(s)

it'd be trivial to alter some strings........

Sure this builds SDK header files are worth a look.

Finally managed to get a screenshot of winshape's About screen:

Wow, very cool linuxlove. How about the video tour?

made an installation ISO without the CHECKED directory in i386i, soon upload

The start of the image is what is now the _IMAGE_FILE_HEADER structure. As louisw3 pointed out, it's begins with the machine type which is IMAGE_FILE_MACHINE_I386. You can also tell the image has 8 sections and was compiled on Wed Oct 16 11:38:29 1991. COMDEX opened just several days later on October 21.

You're right, except that you missed the fact that this is the classic COFF structure. It resembles PE because the PE format was based off COFF.

This is a description of the MIPS COFF Format, but the only relevent difference here is the "magic number" is not the MIPS but the i386 one as you pointed out.

It's a COFF structured file which means it follows the definition of COFF as set forward in the UNIX System V environment but contains further extensions. The format and size used to describe the _FILE_HEADER structure which I mentioned, is that of standard COFF but adapted by Microsoft for Windows NT.

I steered clear of explaining the format in my previous post but instead chose to highlight certain fields. It's wrong to say PE is based on COFF. PE is not based on COFF but uses COFF to describe parts of the PE format. In such case, it would be more correct to say that PE is derived from COFF. PE is for example non architecture specific, COFF is not. I don't doubt you're wrong in your understanding, just wrong in your wording of it.

The colours look a bit like my SNES, But the UI is really weird in them versions of Windows especially the top bar of Windows

That's just how Windows looked prior to Windows 95. I do find it interesting how they're returning to the centered title text in Windows 8 though.