Module state

Structs§

Boundary: A single boundary snapshot (committed or verified).
ChunkTaskInfo: One row from chunk_task for display / debugging.
ExportMetric: One row from export_metrics.
ExportProgression: One export’s progression record.
FileRecord: One row from file_log (formerly file_manifest; renamed in schema v8).
LoadRecord: One rivet load invocation’s ledger record: the load_run audit row plus the extraction source_run_ids it consumed (written into loaded_source_run so a later load skips them). status ∈ success | failed.
MetricRow: Every column written to one export_metrics row.
RunAggregate: One aggregated rivet run.
RunAggregateEntry: Per-export row inside an aggregate.
SchemaChange: Diff between two schema snapshots.
SchemaColumn: One column in a schema snapshot.
ShapeWarning: One column whose observed max byte length grew beyond the configured threshold.
StateStore: Entry point for all persistent state. Supports two backends:

StateRef: Serialisable reference that identifies a state database without holding a live connection. Passed to parallel chunk workers so they can open their own connection for atomic claim_next_chunk_task operations.

arrow_schema_to_columns: Convert an Arrow schema (the dest-facing one, after internal columns are stripped — see pipeline::sink) into the Vec<SchemaColumn> the rest of the trust contract uses (schema_fingerprint, store_schema, detect_schema_change).
schema_fingerprint: Compute the canonical schema fingerprint for a set of columns.