[Lammy]

> The latter four digits could be the time that this build was compiled, which would correspond to 00:22

I'm curious how the Windows build lab was set up in those days. To me something like this says that daily build was started by automation at midnight and took 22 minutes.

> The build functions well on top of standard, standalone MS-DOS for the most part. However, the WinOldAp module that enables running MS-DOS applications under Windows depends on functionality that was implemented by Cougar, the 32-bit DOS kernel project, and is therefore broken under this copy of the build.

Translation: identical outcome as with the explicit https://en.wikipedia.org/wiki/AARD_code but now with plausible legal deniability.

[lloydatkinson]

22 minutes for a full build of an OS written in C in the early 90's seems fast. Then again, it wasn't a big as it is now.

[codezero]

I remember building a circa 1.0 kernel for Linux on a 386 in the early 90s (would have been around 1995) and it took > 24 hours to compile - I'm sure someone could speed run this today, but that's what I remember :)

[codezero]

or the time was UTC, which would put it at ~ 5:22 pm Pacific time (Seattle)

I guess the article says this matches modified times on the files, I wonder if they also have a time zone associated with them.