QueryNodes

Remarks

doc 37 § Part 3's RunQuery, which that table lists twice — once as a task and once as a service — because both forms are wanted and they answer different questions. The task asks now and the branch waits for the answer; the service keeps a key pointed at the best spot for as long as the branch it hangs on is running.

⚠ Two type names, because a schema entry has one slot. RunQuery is the task and KeepQueryResult is the service; calling them both RunQuery would mean the search popup could not tell an author which one they were about to drop, and a file could not say which one it meant.

⚠ A query is named rather than described, and the library is passed in. A compiled query holds live generator and test objects — a PhysicsWorld, a NavMeshQuery — and none of those is a string in a file, which is the same bargain PlaySound and DoesPathExist already make.

Methods (3)

• public static IEnumerable<BehaviorNodeType> Describe()

  The declarations, for a caller adding them to a schema of its own.

• public static BehaviorNodeSchema Register(BehaviorNodeSchema? schema = null)

  Adds them to a schema.

• public static BehaviorTreeResolver Register(BehaviorTreeResolver resolver, EnvironmentQueryLibrary queries)

  Teaches a resolver to build them.