Before arriving to Paris I had no expectations regarding the conference. I was pleasantly surprised by the good event organisation and mostly good talk quality.
The conference kicked off with Roland Kuhn’s keynote. He has done a very good high level overview of the challenges that developers face building large distributed applications and what trade-offs they can make. It’s a talk I highly recommend coming back to for everyone before creating a reactive application architecture.
After the keynote I stayed to listen to what Jon Pretty and Lars Hupel had to say about the Typelevel initiative. It is encouraging to listen how the wide Scala community orients itself on improving the whole ecosystem. Typelevel Scala, which was recently highly publicised - check out [Typelevel's announcement]((http://typelevel.org/blog/2014/09/02/typelevel-scala.html) and Typesafe's response - aims to be a discussion forum and testbed for language features that could later be incorporated into Typesafe Scala. They are looking for contributors.
The next talk I attended was Building a Distributed Data Ingestion System with RabbitMQ by Alvaro Videla. It was rather underwhelming. The speaker at the beginning openly admitted that the talk won’t be related to Scala at all. As much as I think general ecosystem talks are important – I’d expect the speakers to be well prepared on how their product integrates with Scala, Akka and the standard stacks.
Criteo’s talk on stream processing was fun and informative. They explained what is their use case for stream processing, highlighting the amounts of data they ingest (500 billion events a day). What followed was an introduction into the Twitter stack: storm, summingbird, algebird, bijection among others. I was pleased to see a lot of code examples and a good summary of advantages and disadvantages the chosen stack. My key takeaway from the talk is that the Twitter stack is a bit of a lock-in. Once you’re using one tool, you have to use other Twitter tools to get a larger job done.
Fast forward to akka-http talk by Johannes Rudolph and Mathias Doenitz. The speakers presented the reasons behind creation of akka-http, how does it relate to spray and what problems does it solve. Informative talk shedding a lot of light on where akka-http is heading.
The conference day ended with George Leontiev’s hilarious Hanoi tower logic programming implementation in Scala’s type system. There's a Prolog in your Scala! was a fun talk to end the day.
In the evening the conference team prepared something special: a boat trip down the Seine. We toured Paris from the river having food and drinks. A lot of good conversations took place!
I kicked-off the second day, after a slow start, with two talks on data processing. eBay’s Christopher Severs' and Alex Zhicharevich’s From research to production with Scalding, Algebird and Spark and Scalable Genomics with ADAM by Andy Petrella and Xavier Tordoir. The second of the two was especially interesting. Andy and Xavier introduced the audience to challenges of genome sequencing. The most telling slide was one where they showed the decreasing price of this process. As all other components decreased the price of actual data processing stayed almost the same.
Later I attended Implementing the Active Record Pattern in Slick and Beyond URL Matching: Mastering the Spray/Akka-Http Routing DSL. Both talks were very code focused. Age Mooij went even further and had the whole presentation as code (the title slide was a case class construction). Slick and Spray are tools I use very often and it was great to see how other practitioners use them. Both speakers published their code on github: slick talk, spray talk.
All in all the conference was a very fun and enjoyable event. As great event organisation goes: you don’t see it, everything works: food is there when you need it, coffee is decent and WiFi is mostly working – this one went pretty well. With nothing to complain about one could easily focus on the talks. I left Scala.IO with good memories and motivated to learn a few things I heard about.
To sum up Scala.IO gets Highly recommended mark from me.