The need for fast and clear real-time responses are not new to user experience and application development. Instant feedback and system status are core to building better interactive interfaces. Now, our ability to deliver real-time user experiences is becoming more straightforward with the use of LiveView. How can we ensure these features provide the feedback and updates the user is looking for? How can we ensure LiveView delivers the smooth experience we hope for? Applying several common user experience principles, we can bring a base consistency to our real time features while leveraging the power that LiveView brings us to provide live feedback and updates from client interactions AND server-side events.

Scale is a very popular topic among software engineers, and it is something many of us strive to do well, but we rarely discuss its consequences. We think of scale as an inherently good thing. After all, you can make more money, run more efficiently, and solve tough problems. But with this year's theme of building large-scale applications, I'd like to flip that sentiment on its head and make the case against scale.

"If you want to be a great Elixir developers, you need to learn to build your programs in layers, the way the experts do. In this session, we will explore the advice:

Do Fun Things with Big, Loud Workerbees.

This sentence, from Designing Elixir Systems with OTP, is a mental mnemonic that stands for Data, Functions, Tests; Boundaries, Lifecycles and Workers. We'll talk about each of these layers in detail focusing on the role of each independent layer in a self-healing, performant Elixir program"

MIDI is a specification developed in the 80s to allow musical instruments to talk to each other in near real-time. This inter-instrument talk involved sending notes, changing syntesizer patches, and modifying a song's BPM. Since the 80s, MIDI has continued to be the go-to specification for controlling musical instruments, lighting rigs, and more!

In this talk, I'll demonstrate how Elixir can control MIDI devices (specifically, for this talk, an analog synthesizer) in real-time. I'll show how OTP provides a great way to implement this control and the benefits and drawbacks of using it for real-time musical instrumentation.

Furthermore, I will demonstrate how this system is then visualized using Phoenix and LiveView.

“Smart” and connected vehicles generate TONS of data. To make use of that data from a fleet of vehicles for analysis of V2V interactions, traffic patterns, or control flow, you have to consume as much of it as fast as possible with high fidelity. The team at SmartColumbus has spent the last year building robust data ingestion capabilities in Elixir at the speed and scale needed to support thousands of vehicles simultaneously.

This talk will showcase the strategies developed for concurrent processing, batching, and validating data across each of these unique streams. You’ll learn about the design decisions the team has made as well as the open source libraries we’ve developed to enable data consumption at the scale of a “Smart City”.

Scenic is a powerful new library that empowers us to build our own user interfaces from scratch that work outside of the browser. Let’s put it to the test in building our very own video game. We’ll explore a variety of techniques for basic game development such as animation, finite state machines, alternate input devices, etc. With Scenic, is Elixir now the ideal language for game development? Probably not but who cares, let’s just have some fun!

Real time space battle multiplayer browser games are cool. But what if we want to play a real time space battle multiplayer game where the players are hundreds of millions of miles apart, and the speed of light is a factor in who sees what happening, and when?

Hard sci-fi like "The Expanse", or "Dread Empire's Fall" go into excruciating detail on how long it takes to even see what is happening, and how long ago it really did happen - which could be minutes or hours. Battles play out like high latency, async chess games. Warships hurl missiles at each other over enormous distances and speeds, rather than frenetic "World War 2 fighter planes in space" you get in Star Wars.

In a plain old Elixir realtime game, you don't have to worry about that - you just send all messages to all clients as soon as they arrive. But what if we need to delay some of those messages for an arbitrary amount of time depending on where different actors are in the virtual solar system? What design patterns and ways of thinking about how our app is built do we need to do to be able to build something like that?

The problem with a paper airplane, is that it always seems to turn one way or the other. I can never get them to maintain a steady course. What if Nerves was riding along as the pilot?

We'll first talk about inertial measurement sensors and how they can report their position in 3-D space. Then we'll dive into how we can talk to those sensors from a Nerves project. We'll end by talking about some logistics around weight, battery, etc and with any luck a live demo of tossing a glider over the heads of the audience. I'll bring a video as backup in-case the live demo is hazardous or fails.

The Erlang and Elixir embedded software and hardware platform GRiSP is out over 3 years now and now covers Elixir and Erlang with mix and rebar3 tooling and is looking at further places where Erlang VM as a Unikernel might shine.

Join us for a overview of the hardware and software options, how it all works and how our plans for the future shape up.

Elixir, Apache Kafka, and Microservices have all exploded in popularity in the last few years. While their popularities have all exploded, clear understanding of their interplay has not. In this talk, we’ll cover exactly that, and then some. As companies transition from a monolith to a plethora of microservices, asynchronous messaging, the foundation of both Kafka and Elixir, is crucial to well-architected distributed system.

DSC’s Transactional Messaging Service (TMS) is a highly scalable and fault-tolerant Elixir microservice built with Kafka and GenStage and deployed to Kubernetes. It consumes from Kafka to send thousands of transactional messages per minute to our members around the world. We’ll use this service as the lens through which we’ll connect these tools into a truly transformative messaging pipeline.

Writing efficient data structures in Erlang isn't something that's normally done - but by borrowing from both Purely Functional Data Structures, some additional work on Finger Trees, and occasionally cheating with some C code, we can get performance that beats the native OTP implementations. I've written many different versions of data structures for my previous jobs and now I am in the process of bundling up the modules, polishing them up, and open sourcing them for the public's enjoyment and use. Specifically, I'll be covering implementations of mergeable / non-mergeable heaps, queues, and indexable arrays.

As a community we have gone to incredible lengths to avoid ever writing or even looking at Erlang syntax. This has lead to a fractured community, unnecessary dependencies, and opaque code. Even worse, it has prevented us from improving the language that started our ecosystem. This talk points out some common mistakes we make, and ways to improve our code and our ecosystem.