Module graph 

The shared behavior graph model: one serde-friendly data representation every BehaviorInterpreter reads.

A behavior — a behavior tree, a node graph — is authored as a Graph: a set of Nodes, each bound to a function (a statically-known id the interpreter handles natively, or a module call routed by that same id — exactly how arora-behavior-tree already treats its builtins and module actions homogeneously), with typed slots (Io — inputs and outputs, told apart by which Node list holds them) and links (Link) wiring an output/port of one node into the input of another.

This module is only the data model. It does not tick anything and it does not know how any particular interpreter walks the links — the behavior tree reads them as argument/child edges; a node graph reads them as dataflow. Each interpreter lowers this shared model into its own runtime form (see arora-behavior-tree’s graph module for the tree lowering).

Edition happens through GraphDiff: a set of node/link additions and removals (plus predetermined-key overrides). “Loading” a behavior is just Graph::applying a diff onto an Graph::empty graph — which is why BehaviorInterpreter::apply is the one edition entry point.

Structs

Graph

The shared behavior graph: nodes, the links between their slots, the named blackboard variables, and (for tree-shaped interpreters) the root node.

GraphDiff

An edit to a Graph: what to add and remove.

GraphError

A graph edition failed to apply.

Io

A typed slot on a Node — one entry of Node::inputs or Node::outputs. Which of the two lists holds it is what makes it an input or an output; the slot itself carries no direction, so the two can never disagree.

Link

A directed wire feeding one node input from a LinkSource.

Node

A node bound to a function, with its typed inputs/outputs and (optionally) ordered children.

Port

A reference to one slot on one node: (node, port). Generalizes the behavior tree’s NodeParameterId.

Enums

LinkSource

Where the value feeding a Link’s target comes from.