On Sat, Aug 25, 2018 at 06:04:24PM -0700, Lyndon Nerenberg wrote:
Warren Toomey writes:
Hi all, I've just heard that the Usenix
board of directors do not want
to explicitly celebrate the 50th anniversary of Unix.
I had composed a reply and thought it was snarky, but what the heck.
There is a reason I stopped going to Usenix and this is just more
evidence that I made the right choice.
For those who care, here's the back story (and Clem fixed this but long
after I had pulled out).
Somewhere in the 1990's Victor Yodaiken wrote a really cool paper. I was
one of the reviewers and Rob Gingell was as well (Rob was a Sun Fellow
which was the equiv of a VP but in the technical track. Great guy, I had
and have much respect for him).
Victor's paper was about how to do real time and a time sharing system
at the same time. If you know operating systems you know that real time
and time sharing don't mix, people keep trying to mix them and it doesn't
really work. The real time people talk about "good enough" but the people
who know math and operating systems know that you can't truly get both.
Victor got both. He did it by creating a real time kernel that ran Linux,
all of Linux, the kernel and user space, as the real time kernel's idle
process. Linux disabled interrupts, the RT kernel was like, yeah, sure
you did. It was brilliant. It worked. I've seen demos where Linux is
running xperfmon, some other compute load, tarring up /home and untarring
on an NFS server, in other words, beating the crap out of the hardware,
and the RT kernel never missed an event.
Rob was part of the good old boys club that was Unix. He had never heard
of Victor and he rejected the paper while I was a big fan (I believe this
paper was how I met Victor, I sought him out). Rob came up with some noise
about how the RT kernel wasn't POSIX compliant (duh, then it wouldn't be
real time) but the real reason, I believe, was that Victor wasn't part of
the good old boys club.
Roll forward some years and I'm the chair of Linux expo. It was 1998, early
days, but it was apparent that Linux was a thing. And it was a Unix thing
that wasn't part of Usenix. Ellie (or someone, maybe Honeyman, was he part
of the board?) came to me and asked me to pull the Linux folks into Usenix.
They offered me "anything" I wanted, be on the board, be on the review
committee, etc. What I asked for was nothing for me, but something for
everyone, blind peer reviews of paper submissions. They said no and
I've not been to a Usenix conference since. Clem says he cleaned house
but still, it's not blind reviews to this day so far as I know.
And it's a huge shame. The world should have gotten to see Victor's
brilliant have your cake and eat it too real time / time sharing idea.
It didn't because of the old boys club and that's a bummer.
No disrespect to Rob, I think he was just in that space and sort of
knee jerked, he's a good guy, I think he just made a bad call that