SOSP '19 and Halloween
I was at SOSP '19 in Huntsville for the first bit of these week. Deerhurst was an
interesting place for it to be held and I'm glad I didn't have to go extremely far
to get there (compare with holding a conference in Europe or China). This was my
first SOSP and also my first conference and it was a really neat experience. I had
no idea who would be going - so I was surprised when I saw Andy Tanenbaum sitting a
few rows behind me on the bus. I wanted to talk to him but everytime I seemed to run
into him he seemed busy. Some other attendees of note include John Ousterhout (who
seemed pretty unhappy about page limits for final versions of papers), Remzi and
Andrea Arpaci-Dusseau, authors of the OSTEP book (I believe they were both there
but I don't think I ran into Andrea at any point), and Thomas Anderson. I'm definitely
leaving people out.
I had no idea how huge SOSP would be so I was really nervous and excited to see all these huge names. It also turns out that people pronounce it by reading each letter, sometimes so quickly that they would skip the 'O' and just say 'SSP'. I have been saying 'sauce-puh' this entire time. Oh well.
Regarding the contents of the conference, the first session was on machine learning, of course. I'm sure it was all great work but it doesn't interest me all that much. Thankfully there was some interesting discussion on formally verifying the correctness of systems software, which was a topic I was not even aware of - its neat stuff. A team from Google talked about SNAP, which was a microkernel networking approach, sadly closed source as of right now. Some cool talks on the persistence of memory and DRAM, and a historical sort of look at the speed of the linux kernel over the years (and the points where the performance really took some hits).
I met some very interesting people at the conference, heard some very cool talks, spilled lots of coffee on myself, and got a bunch of stickers from the Microsoft table (thanks guys). I was waiting (and hoping) for someone to interject about "GNU + Linux..." but nobody did at any point. One of my favorite points in the conference was during the SIGOPS business meeting - Shan Lu, the SIGOPS chair, asked if we should try and accept more than 40 papers, and before she could even start the straw poll/show of hands, someone shouted 'NO!' I can't comment on the outcome of any of the polls because I had to leave the session a bit early (it was already past 10:30 at this point).
I'm planning on attending SOSP '21 whenever that is. It seems like the whole gang will be there. I was talking to someone about getting Tanenbaum's autograph and he laughed, it seems like there will be plenty of opportunities for me to talk to some of the big names (and waste their time with my questions).
Finally, today is Halloween. I am celebrating by writing a midterm and prepping for a lab tomorrow morning. I might eat some candy tomorrow when it goes on sale.