ANNOUNCE: Akka Streams & HTTP 1.0 MILESTONE 4

[Roland Kuhn]

Dear hakkers,

we are pleased to announce the availability of the fourth milestone of Akka Streams & HTTP. The biggest directly visible feature of this release is a completely new and fluent Java DSL for writing FlowGraphs:

1. final RunnableFlow<Future<List<String>>> result = FlowGraph.factory()
2. .closed(
3. Sink.<List<String>> head(),
4. (builder, out) -> {
5. final UniformFanOutShape<Integer, Integer> bcast = builder.graph(Broadcast.create(2));
6. final UniformFanInShape<Integer, Integer> merge = builder.graph(Merge.create(2));
7.  
8. builder.from(in).via(f1).via(bcast).via(f2).via(merge).via(f3.grouped(1000)).to(out);
9. builder.from(bcast).via(f4).to(merge);
10. });

The biggest internal change shines through this sample as well: the type returned from the FlowGraph factory describes the result of its materialization. In this example we import Sink.head() into the graph as an argument to the factory method, which will propagate the materialized value to the overall graph—materializing and thereby running this graph will return the Future that the head-Sink materializes to. In order to wire this sink up to the rest of the graph, its so-called “shape” (called “out” above) is passed into the block that constructs the graph.

Another thing to notice is that we explicitly import the Merge and Broadcast junctions into the graph, yielding UniformFanInShape and UniformFanOutShape, respectively. This demonstrates that our Graph type is no longer “untyped”, every graph now has a Shape. A Source[T, M] is just a Graph[SourceShape[T], M], which means that it has a shape that has a single output and materializes to a value of type M. Please take a look at the freshly rewritten documentation to see this all in action.

You might be wondering why we did such a big change to the already quite good DSL, in particular one which will break a lot of source code (sorry guys and gals, but you knew what you signed up for when using pre-alpha software in your projects, right? ;-) ). The reason is a deep and foundational one. The central promise and design goal of Akka Streams is “supreme compositionality”, we have been calling stream elements “blueprints that can be freely reused”. When trying to implement these semantics, we discovered that the approach of using the MaterializedMap structure was fundamentally flawed: importing for example a Sink twice would lead to incorrect results when trying to obtain both materialization results, because the user code has only one Key in its hands. The only principled solution to this problem is that materialization results must be combined explicitly when multiple are generated from the same graph, and this principle applies in a hierarchical fashion. For an example see the docs for Java and Scala, here’s an excerpt:

1. import FlowGraph.Implicits._
2. val r: RunnableFlow[(Promise[Unit], Cancellable, Future[Int])] =
3. FlowGraph.closed(source, flow, sink)((_, _, _)) { implicit builder =>
4. (src, f, dst) => src ~> f ~> dst
5. }

The second argument list combines the three values generated by the source, flow and sink into a Tuple3 (using Scala’s lambda short-hand notation, the Java example would use «(src, f, dst) -> new Triplet(src, f, dst)» for a suitably defined Triplet class).

In addition to these fundamental things we have also worked on new features:

• configurable supervision strategy for stream processing stages (PR)
• add ContentEncoding negotiation (thanks André / @2beaucoup !)
• add Referer header to HTTP model
• make ShortNumber path matcher work and add minimum number to repeat modifier
• refactor HTTP RejectionHandler to make it more predictable

For a list of all closed issues, see the milestone on github.

Happy hakking!

Dr. Roland Kuhn
Akka Tech Lead
Typesafe – Reactive apps on the JVM.
twitter: @rolandkuhn

[Giovanni Alberto Caporaletti]

Brilliant! 

Do you still think websockets will be included in 1.0 final?

Thanks for the hard work

G

[Shawn]

Does it support HTTPS yet?

[Giovanni Alberto Caporaletti]

another bug: onPull is not called after absorbTermination().

I openened an issue. Do you have the same problem?
https://github.com/akka/akka/issues/16966

On Friday, 27 February 2015 19:27:41 UTC, Shawn wrote:

Does it support HTTPS yet?

[Quentin Adam]

Thx a lot for the amazing work

Already used by beta feature at Clever Cloud ;-)

Best Regards,