I liked working with Ken.
One fond memory was porting Plan 9 to the SPARC over Christmas break.
People went home, we hacked, Bonnie brought us food. A week later
people came back and we had a new architecture up and running, pretty
much in toto. (The way the system worked made this a joy to do - so
much just worked right away, including for example the ps command.)
One whole day of that was dealing with register windows, which brought
nothing to the table but trouble, and forced us to use R1 as the stack
pointer register. That was not the fondest day.
Architecture people: please just let me pretend your CPU is like a
PDP-11. No surprises, OK? I'll handle the complaints about irrelevance
in the modern age.
-rob
On Thu, Jul 4, 2024 at 12:59 PM segaloco via TUHS <tuhs(a)tuhs.org> wrote:
On Wednesday, July 3rd, 2024 at 6:53 PM, John Levine <johnl(a)taugh.com> wrote:
According to sjenkin(a)canb.auug.org.au:
Developers of Initial Unix arguably were 10x-100x
more productive than IBM OS/360, a ‘best practice’ development at the time,
so what CSRC did differently is worth close examination.
Ken Thompson was an astonishingly productive programmer. I don't think
you can build a business plan that starts with "hire someone like
Ken."
One weekend just for fun he pounded out most of an APL interpreter,
which I then took and spent a month part time adding a few features
like saving and loading workspaces, and adjusting it to use the APL
character set on our funky bitmap terminals at Yale. He did more in
the weekend than I did in the month, and I am not a terrible
programmer.
To add to the praise, Ken, yourself, and others weren't exactly working on modern
115200 baud terminal emulators and IDEs with all the fancy modern tab completion and
automatic linting either. History has given me an appreciation that these sorts of
conveniences work at all. If I'm ever having a really bad day with my tools, I just
think about Ken, Dennis, et. al. hammering away at 33 ASRs making legends happen and
suddenly I don't feel so bad.
- Matt G.