Demo code for the membership team's spike into step functions.
Note: The word 'awesome' in the name refers to the Awesome Product team, not the awesomeness of the code, which, as Roberto pointed out, probably wouldn't justify that title 🐰.
Fire up sbt and type run.
There are two different apps included in this repo, so you will need to launch both of them separately. sbt figures this out and asks you which one you'd like to run. Launch the worker first and the supplier second (to avoid timeout issues).
You can see the result of the executions you create by looking at SimplesWorkerOne in the developer playground Step Functions console (accessible via Janus).
The app in Worker.scala is the important bit in this demo code. It demonstrates how we can use the AWS Java SDK to implement a worker capable of communicating with a Step Functions State Machine, in order to process jobs.
This worker enters and indefinite loop and regularly polls the state machine for work. When there is a task to complete, the state machine sends the worker some JSON as input. When a task is complete, the worker replies to the state machine either with a success message that includes some JSON output, or with a simple failure message.
In the State Machine, activities are blueprints that describe possible tasks that a worker can subscribe to. In this app there is only one activity, but in a more realistic situation there are likely to be many. In this case we may want to create individual polling loops for each activity, and spin these up in different threads.
The app in Supplier.scala is very basic, it's used to send JSON inputs up to state machines in order to kick off executions. These executions are what then provide jobs for the worker.