Where should the type information be: in tags and descriptors 456

Just a data point about what was known at the time.

After the May 17th cancellation of the KL10 followon we had a session with beer and pizza; this was the friday a week and an half later; which must make it May 27th 1983.

After a graveyard beer or five we took a look at what we were using the dec20 for; and what subsbreastutes we would be using.

We took bets on when 1) the first one of us would have access to a system with as much cpu power, disk space and memory for purely personal use 2) when the first one of us would have access to a system as capable in basic OS services (fork, virtual memory, decent file system, protection, emacs, network services, friendly debugger and development system etc etc), for personal use.

The bets ranged from 4 to 15 years for #1 (I bet 8), and 6 to 20 for #2 (i bet 12). We stored a benchmark for reference use.

I was one year off in hardware capabilites, this happened in september 1990, after 7 years and 4 months, and we declared #2 as happened in 1994, after 11 years, based on Linux 1.0 and FreeBSD 2.0

Had we counted Sun's efforts we could have declared it available with the very first Sparc1 and SunOS 3.4, in line with the shortest bets. As luck would have it, none of the participants went to work where Sun systems were in use before 1995.

By early 1984 we had QNX 4.0 as a stopgap measure; a sort-of decent system on stock PC hardware, and we all used that extensively until PC's became capable enough to run BSDs well.

So, when a remote college's IT staff and students were able to see where things were going, what was wrong with the sights of a whole industry?

It was not "just" DEC. DEC was actually among the better of the lot, this is why we took it as hard as we did when they started going under. It was an entire industry, minicomputers that were "out-minied" from the bottom and "out-IBM'ed" from IBN.

But nobody remembers International Harvester either.

-- mrr