To state or not to state?

[giorgosera]

Hi there, I'm super new to Serverless DSL and I have the following question.

In the example 'Event based greeting' there is a single state that is triggered by an event which then executes an action greetingFunction .

I was wondering if the following alternative would be acceptable and if not why?

• Instead of one state I can have two states.
• The first is an Event state which references the greeting event and then immediately transitions to the next state without executing the action.
• Then in the second state (operation state) I execute the action greetingFunction.

I guess my question is when should I group my actions under one state vs spreading them to different states?
e.g. Why would I have an operation state with two actions AFTER an event state when I can put those actions under the event state and get rid of the operation state? I am failing to understand when to choose the one over the other.

It feels like I am misunderstanding something very fundamental so I hope someone can help me!

Thank you

[ricardozanini]

Hi there! Sorry for the long winter I hope you can still see value on my answer.

I guess my question is when should I group my actions under one state vs spreading them to different states?
e.g. Why would I have an operation state with two actions AFTER an event state when I can put those actions under the event state and get rid of the operation state? I am failing to understand when to choose the one over the other.

There's no correct answer. You can group actions in a state when it makes sense to you. For example, when the actions complement each other.

In the new DSL, each task is responsible for a single operation:

do:
  - task1
  - task2
  - task3

In your example, you would have a listen task before calling the greetingFunction.

Check out our example directories to get an idea: https://github.com/serverlessworkflow/specification/tree/main/examples