Geoff Chappell - Software Analyst
Perhaps inevitably, my discovery late last year of a Bug Check From User Mode that persists in the Windows kernel from as far back as Windows NT 3.51 had people asking if it has been there from the very origin of Windows. What about Windows NT 3.1? How should I know? I had never seen a copy. Even among the many discs I retain from MSDN subscriptions in the 1990s, what was then the new Windows that is entirely its own operating system (rather than a large DOS program) goes no further back than Windows NT 3.51. It’s almost as if Microsoft informally disowned earlier versions or at least dismissed them as trial releases.
Now someone has pointed me to a collection of early versions. The provenance is uncertain, but I now have versions 3.10 and 3.50 for inspection. It is fascinating to see what features were born so fully formed that their architecture has not changed in nearly a quarter-century even if they have, of course, been treated to the occasional improvement—and fascinating too, though differently, to see which were plainly marked for immediate reworking.
Many of my pages that explicitly track the archaeology of Windows must yet be revised, and will be, though perhaps with no hurry. For many more, the archaeology is not their reason for existence but thoughts of revision have led me to realise that the pages anyway need more attention than I have yet given them. This applies especially to the annotated catalogues of kernel-mode structures that I started last year. Those that have already changed substantially are listed below, just not as new.