Package com.tnsai.intelligence.rag.vector

Class VectorBackends

java.lang.Object

com.tnsai.intelligence.rag.vector.VectorBackends

public final class VectorBackends
extends Object

Picks at most one optional vector backend for Role bindings.

Qdrant and pgvector can both be installed. A process that sets both URLs is a deployment fault — the user picks by stack.

Since:

0.14.0

Method Summary

Modifier and Type	Method	Description
static Optional<com.tnsai.memory.advanced.VectorIndex>	configured(String scope)	Opens the configured optional backend, or empty for the in-memory store.
static Optional<com.tnsai.memory.advanced.VectorIndex>	configuredManaged(String scope)	Opens a lifecycle-owned backend for the unified retrieval provider.

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Details

configured

public static Optional<com.tnsai.memory.advanced.VectorIndex> configured(String scope)

Opens the configured optional backend, or empty for the in-memory store.

Parameters:

scope - literal binding owner + source-set identity

Returns:

one remote index, or empty

configuredManaged

public static Optional<com.tnsai.memory.advanced.VectorIndex> configuredManaged(String scope)

Opens a lifecycle-owned backend for the unified retrieval provider.

This bridge exists only because the provider and optional backend implementations live in different packages. Direct applications must use configured(String): an index returned here starts a lease heartbeat and is safe only when its VectorIndex.clear() is attached to the owning RetrievalEngine lifecycle.

Parameters:

scope - versioned provider ownership protocol

Returns:

one lifecycle-owned remote index, or empty

Since:

0.15.0