Midwest.IO 2014 Recap

Originally published on the Clarify.io blog. View archived copy.

Earlier this week, I had the priviledge of speaking at the first-ever Midwest.IO conference in Kansas City, Kansas. While I often hit language-specific or even industry-vertical events, this event was completely polyglot including talks ranging from Scala to Haskell to Ruby to Hack and HHVM. There were a number of great talks but there were three that stood out to me personally.

All of the presentations were recorded. Once they’re available, I’ll add the appropriate links here.

First was the opening keynote from Dr Jeff Norris of NASA’s Jet Propulsion Laboratories. He talked about innovation and the ability and willingness to take risk. While he started with examples like Alexander Graham Bell and the telephone, he added his specific experience in relation to the Mars Landers and Rovers. And best of all, he presented from a simple book using an augmented reality rig – using the Vuforia libraries – that included Harry Potter-esque picture animations and 3d renderings of his components. Overall it was informative and entertaining.

 

On the second day, we had Michelle Brush of Cerner discuss some clever compression hacks. She started with a simple black and white image but then expanded it out to include a variety of other data sources with the most complex being earthquake data from Peru. While the compression techniques were not general purpose, the techniques themselves were intriguing. First she uses a quadtree on the simple black and white image but then uses a series of data cleanups to parse the earthquake data. For example, since she only works with 30 days of data, instead of using timestamp data, she uses minute-offsets from 12:00 on the 1st of the month. Further, since the geographic region is limited, instead of starting from 0 degrees for latitude and longitude, she uses offsets that are relevant to the specific data set. By combining each of these techniques, she’s able to compress the data by upwards of 70%. While there are some downsides – for example, needing a custom compression/decompression tool – it’s an interesting tactic when you’re limited on space or bandwidth.

Finally, in the last session of the conference we had Hoop Somuah of 343 Industries talking about building and scaling the backend of Halo 4 on Azure. While I’ve only ever worked with Azure in passing, the concepts and principles he laid out were applicable everywhere. Specifically he talked about Microsoft Research’s project called “Orleans” which is a a framework for cloud computing. Since most components in the cloud are inherently ephemeral, it introduces a completely different mindset that most developers are used to dealing with. By using Orleans, they were able to use an Actor-like model while abstracting away most of the detail into “grains” which are effectively managed Actors. They handle their own creation, destruction, etc.. all you worry about is using them.

The only (minimal) downside of the conference is that with a mostly local crowd, things wrapped early each night. The social the first night closed at 7pm and the conference itself wrapped by 5pm. While it wasn’t a huge problem, it was odd. All of that said, the conference was incredibly well run, all the sessions were recorded, and the staff was super organized in everything they did. I look forward to checking out the conference next year.