Seph

@josephgentle

Hacker. Artist. Part time sellout for hire.

Joined April 2008

Tweets

You blocked @josephgentle

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

  1. 14 Jun 2019
    Undo
  2. 25 May 2019

    this is what I'd talk about if you ran a conference on simplicity. Why aren't game engines / level editors / photoshop collaborative? Because we have to rewrite that code every. damn. time.

    Show this thread
    Undo
  3. 25 May 2019

    We're missing an API for data that changes over time. I think something like this could simplify dozens of existing systems. I can't count how many times I've implemented half baked versions of this sort of thing.

    Show this thread
    Undo
  4. Retweeted

    Thanks so much to everyone who came out to tonight! It was so lovely to meet folks in person. Pictured: the art gallery we collected from folks! Thanks so much , , , and Bruce Shapiro for contributing!

    Undo
  5. 24 Dec 2018
    Show this thread
    Undo
  6. 24 Dec 2018
    Show this thread
    Undo
  7. 23 Dec 2018
    Undo
  8. 28 Oct 2018

    Bahahaha this is a chart of O(n) vs O(log n) algorithms. Red line is the naive memcpy based implementation. Everything else stacks along the bottom of this chart. I'm leaving this here for next time I teach algorithms.

    Show this thread
    Undo
  9. 28 Oct 2018

    Alright one more cool thing: I don't know what data structure an-rope is using, but its getting less efficient over time. In comparison you can see how the skip list in my jumprope library gives much less predictable perf, but is much more stable. 😍 stats!

    Show this thread
    Undo
  10. 28 Oct 2018

    Code: Report generated with `cargo run --release` in the bench/ directory, using rust 1.31.0-nightly.

    Show this thread
    Undo
  11. 28 Oct 2018

    Explorable report here: Comparing random insert / delete operations (weighted to inserts) in a string of size X. Check this out - this is the same test run on a dedicated machine vs my laptop. You can see the laptop thermal throttling in the spread!

    Show this thread
    Undo
  12. 28 Oct 2018

    This is made with criterion, which if its not obvious is *amazing*. I said before rust/etc > JS but npm > all, and that matters more. Now ... . Y'all are kind of killing it

    Show this thread
    Undo
  13. 28 Oct 2018

    Me a couple years ago: Hah there's no efficient implementations of a rope in rust. I'll port my C rope impl over - oh wait, rust is hard. Me now: Hah yeah C code working! Oh wait, there's lots of amazing implementations in cargo. ... They're better than mine, too! O_o

    Show this thread
    Undo
  14. 26 Oct 2018

    If I could snap my fingers and have one problem be magically solved, it would be

    Undo
  15. 25 Oct 2018

    And yeah, perf + security are more complicated. - Can you protect against spectre with (2)? - POSIX processes are a crap security model for apps - In-kernel means we can use a more fine-grained kernel API. We can improve on POSIX perf if syscalls + context switching was cheaper.

    Show this thread
    Undo
  16. 24 Oct 2018

    Native code sandboxing: 1. Run code userland. Use virtual mem + syscalls to isolate the process 2. Compile to wasm and run in-kernel --syscall overhead ++bounds checks. Which would get better perf? Is it app dependant? I'd love to see stats + make the choice myself

    Show this thread
    Undo
  17. 24 Oct 2018

    I'm porting my skiplist based rope implementation to rust. Benchmarking my code vs doing the same thing naively using strings. This is such a beautiful demonstration of O(log n) vs O(n). And you can see how cache size really changes perf at points! It'd be cool to graph this

    Undo
  18. Retweeted
    23 Oct 2018

    Yuko Shimizu, New York based award winning Japanese illustrator

    Undo
  19. 22 Oct 2018

    What is even the argument? Do you hope that bullying wayward artistic children into studying calculus they'll decide to become physicists? That doesn't happen. They'll just end up hating you, hating school and hating mathematics. It is an evil, this thing.

    Show this thread
    Undo
  20. 22 Oct 2018

    I'm amazed how many non-STEM friends I have who get stressed at the mere thought of learning more mathematics. Punishment to students who fail to nurture an interest invokes suffering in this world, with no commensurate benefit. What a sad taint we spread through maths' beauty.

    Show this thread
    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

    ·