Spantree Selects: Strange Loop 2017
Every year since I've joined, Spantree has gone to Strange Loop as a company. It is a wonderful experience that I would highly recommend, but with all the amazing talks, I miss several that interest me. Deciding what to watch the week after is a fun tradition that I outsource to all my coworkers, and here are their thoughts!
Promises and Pitfalls of Persistent Memory Rob Dickinson
Persistent memory is the future, and it is coming very, very soon. No longer will your beautiful functions and data structures have to be mangled to fit into a back-end datastore that speaks SQL or JSON. We are very close to the point where we can all have robust ACID persistence embedded directly in our programming languages with idiomatic data structures. I for one welcome our new persistent overlords.
How and Why we Built our own Column Store Sam Stokes
I love database internals. In this talk, Sam explains how Honeycomb built a custom column-oriented data store for low-latency queries on events with many high-cardinality attributes. What’s noteworthy is how simple it is -- basically a SQL interpreter, Kafka, and clever use of the filesystem. Really elegant engineering.
Just-So Stories for AI: Explaining Black-Box Predictions Sam Ritchie
Machine learning often seems like magic. Sam walks through a few of the the considerations they evaluated between the power of the state of the art Machine Learning algorithms and their understandability. His explanation of how they are trying to leverage their power while keeping the outcome accessible was an interesting combination. Additionally, the conversation around the recent EU General Data Protection Regulation provides an insight into how practitioners are thinking and adjusting to these new regulations.
Alice’s Adventure in Artificial Intelligence-Land Coraline Ada Ehmke
There were several sessions at Strange Loop 2017 that challenged me to think harder about how my skills could be used for the benefit of society. Coraline provides a wonderful narrative of how technology and passion can be combined to create something truly beautiful. Her detailed walk through of the evolution of Alice, a chat bot that has become so much more than that to her companions is intriguing on both a personal and technical level. I’m very excited to see what Coraline and Alice do in the future!
What Happened to Distributed Programing Languages Heather Miller
Distributed Systems aren’t going anywhere, why are we using languages that don’t support distribution by default? An historical tour of distributed programming languages and an excellent summary of the conversations surrounding the idea. This talk keyed me into Erlang/BEAM, which ultimately lead to discovering the Elixir programming language. The talk was originally slated for Papers We Love, but by happy accident found its way to Strange Loop proper. That being said, the talk moves pretty quickly because of time constraints.
How to Play with Deep Space Data Lisa Ballard
Space is amazing. Research vessels like spacecraft, satellites, and probes emit hundreds of amazing images and facts to us every day, but where can we find the data to play around? Lisa demos some cool things happening in the community as well as her contributions to the field. Side note: this talk has an applause break for a gif of a space volcano. We could all use a little more space volcano.
Haxl: A Big Hammer for Concurrency Simon Marlow
This talk showed the power of Haxl, which can allow you to batch, cache, and optimize remote calls. It can also help with debugging by providing the exact steps the user took to encounter the bug and can generate mock data from live requests for use in testing. Haxl leverages the new GHC ApplicativeDo extension, so the talk shows off the usage of that new feature. Toward the end of the talk, it was mentioned that the Scala Fetch library is inspired/based on Haxl. I was having a hard time wrapping my head around what Fetch was for, so this talk was like a two for one for me.
Scaling with Apache Spark (or a lesson in unintended consequences) Holden Karau
You can definitely tell that Holden has used Spark in anger. This stand-up/computer science routine demonstrated several use cases that can cause performance issues in Spark jobs, but liberally sprinkled with cat photos and comedic plugs to buy her book (go buy High Performance Spark!)
How to Hack a Painting Tyler Hobbs
As someone who likes painting and Clojure, this talk was perfect for me. Tyler walks through the considerations and decisions he made to replicate the style and feeling of watercolors for beautiful effects.
Tuning Elasticsearch for English-Language Precision Erin McKean
After using Elasticsearch in several products, Spantree has a set of established practices for searching English text. Erin covered many points I agree with in a concise, informative, and engaging presentation.
Heather Miller's talk, What Happened to Distributed Programing Languages, is unfortunately not available for viewing yet. That should be published in the near future, and we will update the blog.
What Happened to Distributed Programing Languages, Promises and Pitfalls of Persistent Memory, How and Why we Built our own Column Store, and Just-So Stories for AI: Explaining Black-Box Predictions would be our top four picks. Each of these talks had multiple advocates, and are worth checking out!