Struct FinalizeSession 

pub struct FinalizeSession<'a, FR, SR, NC> {
    pub facts: &'a FR,
    pub sessions: &'a SR,
    pub classifier: &'a NC,
    pub confidence_cfg: &'a ConfidenceConfig,
    pub nli_cfg: &'a NliConfig,
    pub merge_cfg: &'a MergeConfig,
}

Borrow-style bundle of every dependency the finalize pipeline needs.

Built inline at the call site (the watcher in Slice-7, or the CLI --finalize trigger), dropped right after FinalizeSession::execute returns. References keep allocation to one borrow per call.

Fields

facts: &'a FR

sessions: &'a SR

classifier: &'a NC

confidence_cfg: &'a ConfidenceConfig

nli_cfg: &'a NliConfig

merge_cfg: &'a MergeConfig

Implementations

impl<'a, FR, SR, NC> FinalizeSession<'a, FR, SR, NC>

where FR: FactRepository, SR: SessionRepository, NC: NliClassifier,

pub async fn execute( &self, session_id: &SessionId, memory_key: &MemoryKey, ) -> Result<FinalizeStats, UseCaseError>

Resolve every pending fact owned by session_id within memory_key.

Ownership is derived from Fact.source_sessions (see module docs): every pending fact whose provenance list contains session_id is in scope. Returns Ok(stats) even on per-fact failures; the only Err paths are store catastrophes that prevent reading the pending or accepted pools.

memory_key scopes the namespace scan. Callers that already know the namespace (the watcher reading SessionState.memory_key(), the CLI with --memory-key) pass it directly; the CLI additionally exposes a discovery fallback that iterates every key when the operator does not name one.