TITLE: Strange Loop 1: Day One AUTHOR: Eugene Wallingford DATE: September 29, 2018 6:19 PM DESC: ----- BODY:
Assembly of Japanese bicycle require great peace of mind.The other was the intellectual breakthrough that unified Phaedrus's philosophy:
Quality is not an object; it is an event.This idea has been on my mind in recent months. It seemed a fitting theme, too, for Strange Loop. On the first day of the conference, I saw mostly a mixture of compiler talks and art talks, including: • @mraleph's "Six Years of Dart", in which he reminisced on the evolution of the language, its ecosystem, and its JIT. I took at least one cool idea from this talk. When he compared the performance of two JITs, he gave a histogram comparing their relative performances, rather than an average improvement. A new system often does better on some programs and worse on others. An average not only loses information; it may mislead. • Jason Dagit's "Your Secrets are Safe with Julia", about a system that explores the use of homomorphic encryption to to compile secure programs. In this context, the key element of security is privacy. As Dagit pointed out, "trust is not transitive", which is especially important when it comes to sharing a person's health data. • I just loved Hannah Davis's talk on "Generating Music From Emotion". She taught me about data sonification and its various forms. She also demonstrated some of her attempts to tease multiple dimensions of human emotion out of large datasets and to use these dimensions to generate music that reflects the data's meaning. Very cool stuff. She also showed the short video Dragon Baby, which made me laugh out loud. • I also really enjoyed "Hackett: A Metaprogrammable Haskell", by Alexis King. I've read about this project on the Racket mailing list for a few years and have long admired King's ability in posts there to present complex ideas clearly and logically. This talk did a great job of explaining that Haskell deserves a powerful macro system like Racket's, that Racket's macro system deserves a powerful type system like Haskell's, and that integrating the two is more challenging than simply adding a stage to the compiler pipeline. I saw two other talks the first day: