Struct PgvectorLongTerm 

pub struct PgvectorLongTerm { /* private fields */ }

LongTermMemory + FilterableLongTermMemory over the single table described in the module docs; cosine recall via the HNSW index, scope enforced by the scope_kind/scope_value columns.

Trait Implementations

impl FilterableLongTermMemory for PgvectorLongTerm

fn recall_filtered<'life0, 'life1, 'life2, 'async_trait>( &'life0 self, scope: Scope, query: &'life1 str, k: usize, candidate_ids: &'life2 [FactId], ) -> Pin<Box<dyn Future<Output = Result<Vec<Fact>, MemoryError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait,

Top-k semantic recall restricted to the given candidate_ids.

fn embedder_id(&self) -> &str

Identifier of the embedder this store was initialised with. Used by recall_filtered_checked() (M2 concrete impls) to hard-fail on cross-embedder queries — re-indexing required before mixing embedder versions.

impl LongTermMemory for PgvectorLongTerm

fn remember<'life0, 'async_trait>( &'life0 self, scope: Scope, fact: Fact, ) -> Pin<Box<dyn Future<Output = Result<FactId, MemoryError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

Store a fact under scope. Returns a stable id.

fn recall<'life0, 'life1, 'async_trait>( &'life0 self, scope: Scope, query: &'life1 str, k: usize, ) -> Pin<Box<dyn Future<Output = Result<Vec<Fact>, MemoryError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait,

Top-k semantic recall under scope for the supplied query.

fn forget<'life0, 'async_trait>( &'life0 self, id: FactId, ) -> Pin<Box<dyn Future<Output = Result<(), MemoryError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

Remove a stored fact.