I had never attended a Scala Days conference before - not even when it was hosted in the city where I live. Shame on me!
This year 2019 marked the 10th yearanniversary of the conference, and I was able to attend thanks to the support of Lunatech, an agency that embodies the Reactive soul in every service that they provide. I have known them for a while as I run the Reactive Amsterdam meetup and they are very active with the community at large. I met them back in March at a meetup where their Xavier Tordoir gave us an introduction to Tensorflow in Scala (this links is to the same talk he gave at another conference) and they offered me the opportunity to tag along to the upcoming Scala Days 2019 in Lausanne. Where can I find the eternal gratitude emoji? 🙂
Attending was a great experience, especially as I am currently other technologies in my day-to-day work. In this post I want to go over my most important tech takeaways, interlaced with random pictures. I wrote more ‘personal’ remarks in a short thread on Twitter.
I had certainly read about GraalVM, but the two talks I attended managed to fully convince me. GraalVM is a universal VM to run code written in different languages. For the JVM family of languages, the most important bit is a different JIT compiler which performs a powerful code analysis that results in advanced optimizations.
The first one was by Vojin Jovanovic, “Run Scala Faster with GraalVM on Any Platform": a benchmark-fueled deck which introduces GraalVM and their offerings (notably the community vs paid supported version).
The second one was a perfect complimentary talk, from Chris Talinger, “Performance Tuning Twitter Services with Graal and ML”. Chris went over the scale of Twitter’s production system, which is simply astounding for its magnitude. The amount of services is in the thousands, while the amount of instances per service goes in the hundreds, for a mass of instances of O(10^5) 😮.
This talk is truly inspiring and funny (plus you hear me laugh loudly at 55:13). It turns out that the GraalVM optimisations are very effective with Scala code as Scala has lots of constructs and hidden object instantiations. It turns out that the classic C1 and C2 JIT compilers were not written with other languages than Java in mind, which is kinda obvious as most of them did not exist at the time. Twitter runs its services in its own data centers: little tweaks and optimisation can have a tremendous impact on costs.
If you are using Java 11, enabling GraalVM is as simple as adding three compilation options:
java [your other options] -XX:+UnlockExperimentalVMOptions \ -XX:+EnableJVMCI \ -XX:+UseJVMCICompiler
Picture below: enjoying the real Swiss fondue with the Lunatech crew!
This does not come as a surprise. I am a huge fan of Akka Streams and have used it for many projects, both personally and professionally. I have given talks and workshops about it. I am currently working on a .NET environment and I find nothing comparable in terms of flexibility, power and ease of use. The talk I attended by Heiko Seeberger, “Akka Streams to the Extreme”, was a good way for me to catch-up on the latest best practices and new features about this great piece of technology. Check it out!
Picture below: Lunatech organised a support funding for the Let Luna Shine cause, and they would donate 1 euro for each picture taken with their mascotte.
Martin Oderski opened the conference (who else could have done it?) and his keynote “A Tour of Scala 3”
went over the features that he believes will bring the largest benefits to Scala practitioners more. Enums, union types,
a rework of the concept of implicits (future
delegates) to name a few. I am very much looking forward to this change, and
there is no risk of ‘Python syndrome’, mostly thanks to the fact that Scala is strongly typed! ✌️
I hope that Scala 3 will contribute positively to its adoption, but my heart is somehow a little doubtful. I do not see people who would not consider the language earlier now embracing it because of union types or delegates. Maybe it is just me not seeing an ocean of possibilities. To me Scala is already a fantastic tool with an unparalleled ecosystem, so I welcome any improvement. My question is more if the new features will be impressive for people who have not turned to the light yet 😇 In fact, I believe that for the average Joe (like me), the ecosystem around Scala is a bigger selling point than the language features themselves. We will see what happens! Our hearts are with Scala.
I would have liked to highlight more content and probably I have missed lots of great content too - that happens when you have four parallel tracks!