Tweets

You blocked @sadisticsystems

Are you sure you want to view these Tweets? Viewing Tweets won't unblock @sadisticsystems

  1. Pinned Tweet
    Apr 24

    Started an effort I'm calling SLEDSAFE for drilling HARD into some correctness constraints that causally control critical emergent properties: * no data loss * linearizable single-key operations * serializable transactions * max (1 writer) space amp of 3

    Undo
  2. Retweeted
    24 hours ago
    Replying to

    On the relationship between epidemics and periods of crisis in European history. R. Evans, Epidemics and Revolutions (1988)

    Undo
  3. May 5

    Over time I mix more and more testing code into my implementations. Asserts, random delay injectors, random error injectors, and over time I use more and more property tests that inject failures and delays deterministically. Unit/integration tests don't find nearly as many bugs.

    Undo
  4. May 1

    If you like Engineering a Safer World, here are 2 books that really drill into STPA & CAST targeted at practitioners. I started applying STPA for hazard analysis for and find these pretty thorough & helpful

    Undo
  5. May 1

    This has a distorting effect on how conflict is treated in the community. People act like conflict is bad. Healthy normal conflicts turn into terrible persistent situations that can't get solved because they aren't supposed to happen. It's inhuman.

    Show this thread
    Undo
  6. May 1

    Stockholm syndrome among people who got jobs at terrible companies is so real. People will excuse the most disgusting behavior because they don't want to risk a job that lets them write Rust despite feeling like shit because of the job.

    Show this thread
    Undo
  7. May 1

    perf is really good, esp new versions built from the linux git repo in tools/perf `perf list` shows you all the things you can correlate to code. cache misses, branching, so many things `perf report`: interactive TUI of results. Hotspot is also great.

    Undo
  8. Apr 30

    broke: minimizing CPU instructions for your code on some webpage woke: minimizing cache misses on real HW with real workloads cache rules everything. instructions basically don't matter compared to your access patterns.

    Undo
  9. Apr 27

    Although it is quite frustrating that many of the papers are paywalled :/

    Show this thread
    Undo
  10. Apr 27

    While I'm still bummed about not going to Crete this year for papoc & eurosys, it is pretty nice that I can be on the zoom call for free without an academic affiliation and watch it in my swimsuit getting sun on my balcony :)

    Show this thread
    Undo
  11. Retweeted
    Apr 27

    New paper out today! This morning I will be presenting our latest paper "Paxos vs Raft: Have we reached consensus on distributed consensus" at the PaPoC workshop.

    Show this thread
    Undo
  12. Apr 26

    Typing notifications are honestly the most anxiety-producing feature I wish I could disable in every single chat app I use.

    Undo
  13. Apr 24

    This has turned out to be a bigger deal for sled users than I imagined it might be. If you write libraries in , you may be doing your users a bigger service than you realize by supporting a specific version of Rust and up! I chose 1.37 but anything clear is helpful.

    Undo
  14. Apr 24

    It was surprisingly easy to migrate sled to a minimum-supported Rust version of 1.37. We now have a MSRV :) Just had to avoid mem::take and work around MaybeUninit's missing Debug impl. 2020 is the year when sled becomes stable-as-fuck!

    Undo
  15. Retweeted
    Apr 21

    Here's our take on 's new child, io_uring. tl;dr: it's great!

    Undo
  16. Apr 21

    does anyone else read technical books from the back of the book toward the front? I feel like it's the only way I can keep myself from getting pulled into progressions that don't really relate to what I'm opening the book to learn in the first place.

    Undo
  17. Apr 21

    NEW BOOK DAY :D This one also counts as weight training

    Undo
  18. Retweeted
    Apr 20

    Everyone who builds high performance systems should try to answer the three polls. How you structure your measurements radically alters what numbers you'll see, and it's extremely counterintuitive if you don't think long and hard about what you're measuring.

    Undo
  19. Apr 20

    PaPoC is looking really great this year! You can sign up via

    Undo
  20. Apr 20
    Undo

Loading seems to be taking a while.

Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.

    You may also like

    ·