sploreg / goap Goto Github PK

Thank you for the great examples! I am building a Rimworld/Prison Architect type game and GOAP is just what I was needing.

I see in line 135 of GoapAgent.cs as shown below code, but there is else condition at 170 which solves a similar problem. can you explain in which condition this is needed?
if (!hasActionPlan()) { // no actions to perform Debug.Log("<color=red>Done actions</color>"); fsm.popState(); fsm.pushState(idleState); dataProvider.actionsFinished(); return; }

I would suggest going backwards makes more sense when pathfinding to endState.
https://github.com/sploreg/goap/blob/master/Assets/Standard%20Assets/Scripts/AI/GOAP/GoapPlanner.cs#L93

That way you give the AI a higher chance of finding a solution that works instead of a predefined set of actions that give you the result.
In other words, you should match endState with postEffects and the action's postEffects with next actions preConditions or existing startState.

endState -> postEffects -> preConditions/startState
Though you want to make sure to always check against the startState for every action because you could end the planning earlier since it might just be sufficient enough from where you are.

// if the key exists in the current state, update the Value
bool exists = false;

foreach (KeyValuePair<string,object> s in state) {
	if (s.Equals(change)) {  // <-- problem?
		exists = true;
		break;
	}
}

am I missing something? This appears to be comparing whole KVPs, not just the keys.

if (s.Key.Equals(change.Key)) {

appears to work.