rodoherty1 / AkkaStreams

Collection of tests and apps to demo basic functionality of Akka Streams

AkkaStreams

Basic Examples in Unit Tests

• Sources, Folds and Sinks
• Materialized Values from Sources and Sinks
• Tickers
• Simple Websocket Client and Server

Threading in Unit Tests

• Demonstrate that an Akka Stream is not called on the caller's thread
• Demomstrate that AkkaStreams will generally reuse a thread when executing successive processing stages
• Demomstrate that AkkaStreams will occasionally suspend a thread and use a different thread from the threadpool.

Akka-Http

• Akka-Http was intended to be a primary use case for AkkaStreams
• Akka-Http provide back pressure all the way back to the TCP layer

Links

• [Akka Stream docs] (http://doc.akka.io/docs/akka/current/scala/stream/stream-integrations.html)
• [Akka Streams Terminology] (http://doc.akka.io/docs/akka/2.5.2/java/stream/stream-flows-and-basics.html)
• [Integrating Akka Streams and Actors - Colin Breck] (http://blog.colinbreck.com/integrating-akka-streams-and-akka-actors-part-i)
• [Actors in an Akka Stream] (https://stackoverflow.com/questions/39125760/creating-a-flow-from-actor-in-akka-streams)
• [Threading in AkkaStreams] (http://akka.io/blog/2016/07/06/threading-and-concurrency-in-akka-streams-explained)
• [Heiko Seeberger's talk on AkkaHttp and Graphs] (https://www.youtube.com/watch?v=ryxrWVI3PMA&t=1191s)
• [Graph Stages 1] (http://akka.io/blog/2016/07/30/mastering-graph-stage-part-1)
• [Graph Stages 2] (http://akka.io/blog/2016/08/25/simple-sink-source-with-graphstage)
• [Akka's Akka Stream Announcement] (https://www.lightbend.com/blog/typesafe-announces-akka-streams)

Topics that are not covered in this session

• Supervision strategies
• Streams insides Streams (e.g. an ActorRef that kicks off its own Stream)
• Distributed Streams
• The Graph DSL and combining Partial Graphs

FAQ

Q: Should I use many ActorMaterializers? A: You should not need more than a couple of Materializers and those cases where you either want

1. to bind the lifecycle of a bunch of streams to an Actor (in which case you create a Materializer with the Actor's context.system so that when the actor dies, all streams under that materializer will be terminated).
2. if you want to shut down a bunch of streams together by calling materializer.shutdown() which will terminate all streams belonging to that materializer.

Must haves Todos

• Document the basic terminology (graph stages, processing stage)
• Tidy up the websock server and client example. Describe the streams on either side and describe how backpressure works.