Monix Client/Server Sample
Sample of a mixed client/server application demonstrating the streaming of values over web-socket, both simple and with back-pressure applied (by means of the Reactive Streams protocol), using Monix for both the server and the client ;-)
See: monix-sample.herokuapp.com
Overview of the code:
- the server-side values generator is in engine.DataProducer
- the server-side WebSocket channel that implements the back-pressure protocol as defined by Reactive Streams is handled by engine.BackPressuredWebSocketActor (using Play's WebSocket support) and the client-side consumer is client.BackPressuredWebSocketClient
- the server-side WebSocket channel that doesn't do back-pressure is engine.SimpleWebSocketActor while the client-side consumer is client.SimpleWebSocketClient
- the type-safe observable that listens to a server web-socket connection generating signals is in client.DataConsumer
- the integration with Epoch, our charting library, is in client.Graph
NOTES:
- we are exposing 2 versions of webSocket connections, one that is back-pressured as a matter of the server-side protocol and one that is not
- both versions are protected by buffers that start dropping events in case the client is too slow, but the difference is that for the back-pressured version the buffer is being maintained server-side
- by applying back-pressure in the protocol, the server is informed of the rate at which the client can consume and thus there is no risk for the server in case we've got clients that are too slow; on the other hand for the simple version the server can be crashed on clients that are too slow in receiving their events, as that actor's mailbox is unbounded
- the back-pressured version needs server-side cooperation / implementation and is thus more difficult to develop
In order to develop and execute the project locally:
sbt run
For Heroku deployment:
sbt stage deployHeroku
The project is based on the cool play-with-scalajs-example template by @vmunier, so you get the auto-reload coolness of Play in combination with Scala.js.
~;Enjoy