Group meeting

Introduction to wildcard actors and conditions

The previous tutorial had our two actors introducing themselves. In that scenario, we've given one actor the role of initiator and another of recipient.

In this tutorial, we'll model a situation where a random number of people have a conversation. They first all need to introduce themselves after which each one can talk.

  • Create file meeting.yaml in scenarios.

  • Create subdirectory meeting in features for the test files.

Wildcard actors

In this scenario, we don't exactly know how many participants the process will have. However, we do that each participant has essentially the same role. We'll number the actors when we instantiate the process.

Test case

In the initial state, everyone has the opportunity to introduce themselves. With an introduction, we stay in the initial state. When someone starts to talk we transition to the talking state.

meeting/golden.feature
Feature: Three persons introduce themselves and have a conversation

  Background:
    Given the process is created from the "meeting" scenario
    And "Bob" is the "person_1" actor
    And "Eve" is the "person_2" actor
    And "Jack" is the "person_3" actor

  Scenario:
    When "Bob" does "introduce" with:
      | name         | Bob   |
      | organization | Jasny |
    Then the process is in "initial"
    And the last event is not skipped
    And actor "person_1" has "name" is "Bob"
    And actor "person_1" has "organization" is "Jasny"

    When "Eve" does "introduce" with:
      | name         | Eve      |
      | organization | Acme Inc |
    Then the process is in "initial"
    And the last event is not skipped
    And actor "person_2" has "name" is "Eve"
    And actor "person_2" has "organization" is "Acme Inc"

    When "Jack" does "introduce" with:
      | name         | Jack        |
      | organization | LTO Network |
    Then the process is in "initial"
    And the last event is not skipped
    And actor "person_3" has "name" is "Jack"
    And actor "person_3" has "organization" is "LTO Network"

    When "Bob" does "talk" with "What do you think about the weather?"
    Then the process is in "talking"
    And the last event is not skipped

    When "Eve" does "talk" with "It's pretty cold"
    Then the process is in "talking"
    And the last event is not skipped

    When "Bob" does "talk" with "True, but at least it's not raining"
    Then the process is in "talking"
    And the last event is not skipped

    When "Jack" does "talk" with "I like blockchain!"
    Then the process is in "talking"
    And the last event is not skipped
    
    Then the result is:
      """
      - Bob - What do you think about the weather?
      - Eve - It's pretty cold
      - Bob - True, but at least it's not raining
      - Jack - I like blockchain!
      """

The scenario

Instead of defining the actors individually, we use a wildcard. The actor schema is applied to every actor with the key person_{number}.

For goto we can use null (which is ~ in YAML) to indicate that we don't want to transition to another state.

meeting.yaml
actors:
  person_*:
    properties:
      name: string
      organization: string    

actions:
  introduce:
    response:
      type: object
      properties:
        name: !required string
        organization: !required string  
    update:
      set: current.actor
      mode: merge
  talk:
    response: string
    update:
      set: result
      mode: append
      value: !tpl '{{ current.actor.name }} - {{ current.response }}'

states:
  initial:
    transitions:
      - on: introduce
        goto: ~
      - on: talk
        goto: talking
  talking:
    transitions:
      - on: talk
        goto: ~

result:
  type: array
  items: string

It's bad practice to have a process without an end state. We'll add a way to end the conversation later in this tutorial.

Everyone introduced?

With the scenario of the introduction tutorial, the state machine made sure that both parties did an introduction before being able to start the conversation. However, with this scenario that's not the case. Any actor can start talking and skip the introductions. Additionally, we don't need actors to introduce themselves twice.

Test case

meeting/conditions.feature
Feature: People are not committing to the social norms

  Background:
    Given the process is created from the "meeting" scenario
    And "Bob" is the "person_1" actor
    And "Eve" is the "person_2" actor
    And "Jack" is the "person_3" actor

  Scenario: Person introduces himself and starts to talk
    When "Bob" does "introduce" with:
      | name         | Bob   |
      | organization | Jasny |
    And "Bob" does "talk" with "What do you think about the weather?"
    Then the last event is skipped

  Scenario: Person starts talking without introducing himself
    When "Bob" does "introduce" with:
      | name         | Bob   |
      | organization | Jasny |
    When "Eve" does "introduce" with:
      | name         | Eve      |
      | organization | Acme Inc |
    When "Jack" does "talk" with "I like blockchain!"
    Then the last event is skipped
    
  Scenario: Person does an introduction twice
    When "Bob" does "introduce" with:
      | name         | Bob   |
      | organization | Jasny |
    When "Bob" does "introduce" with:
      | name         | Bob Builder |
      | organization | Jasny       |
    Then the last event is skipped
    
  Scenario: Person tries to talk twice
    When "Bob" does "introduce" with:
      | name         | Bob   |
      | organization | Jasny |
    When "Eve" does "introduce" with:
      | name         | Eve      |
      | organization | Acme Inc |
    When "Jack" does "introduce" with:
      | name         | Jack        |
      | organization | LTO Network |
    When "Bob" does "talk"
    Then the process is in "wait_on_recipient"
    When "Bob" does "talk"
    Then the process is in "wait_on_recipient"
    And the last event is skipped

The scenario

We can use if conditions to disallow an action or a transition. The field needs to contain a boolean value. We typically use <ref> data function for conditions. This function performs a JMESPath query against the process.

For the "introduce" action we specify a condition to prevent an actor from performing it if we already know its name.

We'll add a condition statement for state transition from "initial" to "talking", which validates that the name of every actor is known.

The last condition is on the "talk" action. We compare if the current actor is not the actor of the previous event to ensure a person doesn't talk twice.

meeting.yaml
actors:
  person_*:
    properties:
      name: string
      organization: string

actions:
  introduce:
    if: !ref current.actor.name == null
    response:
      type: object
      properties:
        name: !required string
        organization: !required string
    update:
      set: current.actor
      mode: merge
  talk:
    if: !ref events[-1].actor.key != (events[?(!skipped && actor)] | @[-2]).actor.key
    response: string
    update:
      set: result
      mode: append
      value: !tpl '{{ current.actor.name }} - {{ current.response }}'

states:
  initial:
    transitions:
      - on: introduce
        goto: ~
      - on: talk
        if: !ref values(actors)[?name == null] | length(@) == 0
        goto: talking
  talking:
    transitions:
      - on: talk
        goto: ~
      - on: end
        goto: (done)

result:
  type: array
  items: string

JMESPath explained

current.actor is a reference to the object of actors. That object doesn't contain the actor key. Instead, we need to get the value from the events.

The last event is the event of the current action, with JMESPath that's events[-1].

The previous event would be events[-2]. However, we want to exclude timeout and skipped events, so we do (events[?(!skipped && actor)] | @[-2]).

No introductions

In case the persons already know each other, there is no need for introductions. We can initialize actors using the start instructions of the process.

meeting/no-introductions.feature
Feature: Two persons have a conversation without an introduction

  Background:
    Given the process is created from the "meeting" scenario
    And "Bob" is the "person_1" actor with:
      | name         | Bob   |
      | organization | Jasny |
    And "Eve" is the "person_2" actor with:
      | name         | Eve      |
      | organization | Acme Inc |

  Scenario:
    When "Bob" does "talk" with "What do you think about the weather?"
    Then the process is in "talking"
    And the last event is not skipped

    When "Eve" does "talk" with "It's pretty cold"
    Then the process is in "talking"
    And the last event is not skipped
    
    Then the result is:
      """
      - Bob - What do you think about the weather?
      - Eve - It's pretty cold
      """

Leaving the conversation

Instead of ending the interaction, any person can choose to leave. The remaining persons can continue the conversation. When there are less than two persons left, the process should end.

Test case

meeting/leave.feature
Feature: People leave the conversation

  Background:
    Given the process is created from the "meeting" scenario
    And "Bob" is the "person_1" actor with:
      | name         | Bob   |
      | organization | Jasny |
    And "Eve" is the "person_2" actor with:
      | name         | Eve      |
      | organization | Acme Inc |
    And "Jack" is the "person_3" actor with:
      | name         | Jack        |
      | organization | LTO Network |
    Then actor "person_1" has "is_present" is true
    And actor "person_2" has "is_present" is true
    And actor "person_3" has "is_present" is true

  Scenario: Person leaves and later the second leaves
    When "Bob" does "talk" with "What do you think about the weather?"
    When "Eve" does "talk" with "It's pretty cold"
    When "Bob" does "talk" with "True, but at least it's not raining"
    When "Jack" does "talk" with "I like blockchain!"

    When "Bob" does "leave"
    Then actor "person_1" has "is_present" is false
    And the process is in "talking"

    When "Eve" does "talk" with "What is blockchain?"
    And "Jack" does "talk" with "Do you know Bitcoin?"
    And "Eve" does "talk" with "*sigh*"

    When "Eve" does "leave"
    Then actor "person_2" has "is_present" is false
    And the process ended with:
      """
      - Bob - What do you think about the weather?
      - Eve - It's pretty cold
      - Bob - True, but at least it's not raining
      - Jack - I like blockchain!
      - Eve - What is blockchain?
      - Jack - Do you know Bitcoin?
      - Eve - *sigh*
      """

    Scenario: Person tries to say something after already left
      When "Bob" does "talk" with "What do you think about the weather?"
      And "Eve" does "talk" with "It's pretty cold"

      When "Bob" does "leave"
      And "Bob" does "talk" with "True, but at least it's not raining"
      Then the last event is skipped

The scenario

We add an is_present property to the actor, which defaults to true. The "leave" action will set is_present to false for the current actor.

As long as there are two or more actors present, the "leave" action will not cause a state transition. If that condition is not true, the process will end. Transitions are validated in order. Therefore we only need to have a condition on the first transition of "leave" action in the "talking" state.

meeting.yaml
actors:
  person_*:
    properties:
      name: string
      organization: string
      is_present: !default true

actions:
  introduce:
    if: !ref current.actor.name == null
    response:
      type: object
      properties:
        name: !required string
        organization: !required string
    update:
      set: current.actor
      mode: merge
  talk:
    if: !ref current.actor.is_present && (events[-1].actor.key != (events[?(!skipped && actor)] | @[-2]).actor.key)
    response: string
    update:
      set: result
      mode: append
      value: !tpl '{{ current.actor.name }} - {{ current.response }}'
  leave:
    if: !ref current.actor.is_present
    update:
      set: current.actor.is_present
      value: false

states:
  initial:
    transitions:
      - on: introduce
        goto: ~
      - on: talk
        if: !ref values(actors)[?name == null] | length(@) == 0
        goto: talking
  talking:
    transitions:
      - on: talk
        goto: ~
      - on: leave
        if: !ref values(actors)[?is_present] | length(@) >= 2
        goto: ~
      - on: leave
        goto: (done)

result:
  type: array
  items: string
PreviousA proper introductionNextAuthentication

Last updated