Module daemon_graph 

In-process graph backing store for the loopback daemon /query route.

The live daemon graph is a fixed-capacity, zero-heap store backed by a caller-invisible [NQuin; MAX_GRAPH_QUINS] buffer. Cold-path ontology and file ingestion may still allocate while parsing, but the resident graph used by /query no longer relies on Vec or HashSet.

Structs

DaemonGraphStore

InitGraphOptions

Controls which resident graph layers are seeded at daemon boot.

UpdateOutcome

Outcome of applying a SPARQL Update to the daemon graph.

Constants

MAX_GRAPH_QUINS

Bench datasets (Schema.org ~18K quins) must fit for browser/native parity.

Functions

apply_sparql_update

Apply a parsed SPARQL Update to the resident daemon graph.

apply_sparql_update_durable

Apply a SPARQL Update durably and signed: each net change is stamped and signed to the WAL (the append-only audit trail) with the caller’s real ed25519 key, then the full graph state is snapshotted for restart durability. This is the production write path; the key must come from the identity / key-vault layer — never a placeholder.

bump_graph_revision

Increment revision and notify SSE subscribers (Release ordering).

condition_label_for_subject_hash

Known condition subject hashes for Anatomy graph -> label mapping.

extend_with_ontology_quins

Extend the live graph with ontology quins from qualia-core-db::ontology_loader.

extend_with_ontology_quins_slice

Zero-heap resident update path for ontology insertion.

graph_lexicon_lookup

Resolve a term hash to its literal text via the resident graph’s lexicon. Used by the query path as the ingested-data resolver for TextResolver.

graph_quin_count

Number of Quins currently available to /query.

graph_read_guard

Read guard over the live graph (lock is process-static via OnceLock).

graph_revision

Monotonic Lamport-style revision counter for daemon graph mutations.

init_daemon_graph

Initialise or refresh the daemon graph from storage path.

init_daemon_graph_with_options

Initialise or refresh the daemon graph with explicit seeding policy.

load_graph_snapshot

Load the durable snapshot into the graph, if present. Returns the quin count loaded (0 if there is no snapshot). This is the authoritative last full state (defaults + index + committed updates) when it exists.

persist_graph_snapshot

Write the current full graph state to a flat-quin snapshot file. Called after a persisted SPARQL Update so the change survives a restart. Returns the quin count written.

replace_graph_from_flat_bytes

Replace the in-memory graph with flat 48-byte NQuin bytes (browser bench_load).

subscribe_graph_revisions

Subscribe to graph revision bumps (used by GET /tensor/events SSE).