API Reference — dna/layers/MachineLearnedBehaviorExtWriter

---

class MachineLearnedBehaviorExtWriter : public virtual MachineLearnedBehaviorWriter

Populate and clear the ML extension operation data stored in a rig's DNA layer.

When to use this

Use this interface when implementing a custom Writer subclass that needs to write ML extended behavior operation sets into a DNA asset. It provides the full mutation surface for operation type assignment, static parameter injection, and dependency graph construction across the three-level hierarchy of ML type → operation set → operation.

Watch out for

• Do not inherit from MachineLearnedBehaviorExtWriter directly. The @warning on the class requires that implementors inherit from Writer itself. This class exists solely to group the extended ML write accessors; direct inheritance bypasses the full writer contract.

Method groups

Group	Methods
Bulk clear	clearMLExtData, clearMLOperationSets, clearMLOperations
Set operation data	setMLOperationType, setMLOperationParameters

Example

// Concrete Writer subclass (inherit from Writer, not this class directly)
class MyRigWriter : public Writer {
public:
    void writeMLExtensions(MachineLearnedBehaviorExtWriter* w,
                           std::uint16_t mlTypeIndex) {
        // Clear any prior state for this ML type before repopulating
        w->clearMLOperationSets(mlTypeIndex);

        // Assign the operation type for operation 0 in set 0
        w->setMLOperationType(mlTypeIndex, /*setIndex=*/0, /*opIndex=*/0,
                              MachineLearnedBehaviorOperationType::Multiply);

        // Attach static parameters to the same operation
        const std::uint32_t params[] = {1024u, 2u};
        w->setMLOperationParameters(mlTypeIndex, 0, 0, params, 2);
    }
};

Parameters

All mutating methods share the same three-level index scheme:

Name	Type	Description
mlTypeIndex	std::uint16_t	required — zero-based index of the ML model in the rig's ML model array
mlOperationSetIndex	std::uint16_t	required — zero-based index of the operation set within the selected ML model
mlOperationIndex	std::uint16_t	required — zero-based index of the individual operation within the selected set
operationType	MachineLearnedBehaviorOperationType	required (setMLOperationType only) — the operation variant to assign
parameters	const std::uint32_t*	required (setMLOperationParameters only) — pointer to the static parameter array to copy
count	std::uint16_t	required (setMLOperationParameters only) — number of elements to copy from parameters

Constraints

• Storage at all three levels is implicitly resized when a set* call references an index beyond the current array bounds. No pre-allocation step is required, but callers should write indices in a predictable order to avoid sparse allocations.
• Clearing is hierarchical: clearMLExtData removes all ML extension data; clearMLOperationSets removes all sets for one ML type; clearMLOperations removes all operations within one set.