Module frozen 

Immutable dictionary-encoded RDF dataset built at the inference→validation boundary. Implements spareval::QueryableDataset so Oxigraph’s prepared SPARQL evaluator can execute queries against it without a mutable Store.

Stage 2 of docs/05-sparql-execution.md.

This file is the read-mostly dataset boundary used after inference. Loading assigns compact TermIds once, then builds three sorted triple permutations for the prefix scans that paths and native SPARQL need. Evaluation stays in ids until a result crosses back into RDF terms; avoiding repeated Term hashing and cloning is the point of the representation.

Query-only terms and reachability results may be cached behind interior mutability, but neither changes the RDF snapshot. The only real mutation, extend_triples, updates every index/statistic and discards derived closures. Thus every scan sees one coherent snapshot and every cached closure belongs to that snapshot.

Structs

DatasetStatistics

Dataset statistics used in stage 3+ for join-order planning (doc §198).

FrozenIndexedDataset

Immutable, dictionary-encoded snapshot of a post-inference RDF dataset. Intended to be built once at the inference→validation boundary and shared across all per-focus-node SPARQL evaluations.

Type Aliases

TermId

Dictionary index for a term. Chosen as u32 to keep triple arrays at 12 bytes each; the 4B ceiling is unreachable in practice.