Fix event transition to "Complete" creation and trigger in esmini for events containing ParameterSetAction

[sumanth-kandala-ai]

Closes #677

Context:

Details

When defining an Event that contains a single Action which is only responsible for setting a parameter like shown in the snippet below:

<Event maximumExecutionCount="1" name="Event 1" priority="parallel">
  <Action name="Action 1">
  	<GlobalAction>
  		<ParameterAction parameterRef="my_parameter">
  			<SetAction value="7" />
  		</ParameterAction>
  	</GlobalAction>
  </Action>
  <StartTrigger>
      <ConditionGroup>
          <Condition delay="0" conditionEdge="none" name="simulation_time">
              <ByValueCondition>
                  <SimulationTimeCondition value="0.0" rule="greaterThan"/>
              </ByValueCondition>
          </Condition>
      </ConditionGroup>
  </StartTrigger>
</Event>

The following logs show up on running a scenario with the above pattern:

[0.000] [info] Event 1 Condition 1-1: true, delay: 0.00, 0.0000 >= 0.0000, edge: none
[0.000] [info] Trigger /------------------------------------------------
[0.000] [info] Group 0:
[0.000] [info] Event 1 Condition 1-1: true
[0.000] [info] Trigger  ------------------------------------------------/
[0.000] [info] Event 1 standbyState -> startTransition -> runningState
[0.000] [info] Set parameter my_parameter = 1
[0.000] [info] Event 1 Action 1 initState -> startTransition -> runningState
[0.000] [info] Event 1 Action 1 runningState -> stopTransition -> completeState

Interestingly we don't see the logs at 0 seconds which blocks any Events defined subsequently:

[0.000] [info] Event 1 complete after 1 execution
[0.000] [info] Event 1 runningState -> endTransition -> completeState

Furthermore this same pattern used to work in older versions of esmini, i.e. esmini-2.36.2 and below.

Root Cause: It seems like this commit caused this regression.

Specifically in ScenarioReader.cpp the following change was made:

-                     ParameterSetAction *paramSetAction = new ParameterSetAction();
+                    ParameterSetAction *paramSetAction = new ParameterSetAction(action);

In contrast VariableSetAction, which works on the current version of esmini uses parent too instead of action and seems to work as expected. This also seems consistent since the empty constructor for ParameterSetAction in the previously working version of esmini here is now equivalent to the constructor that takes in a parent and not an action.

Let me know if you'd like me to add any tests for the same.

[eknabevcc]

Wonderful. Nice find and fix, indeed! Thanks.

[eknabevcc]

Included in release v2.46.0