Dataset, Source, and Snapshot

pChronicle is an Agent trajectory storage engine. A Dataset is a path: the physical origin of trajectory data stays visible instead of hiding every trajectory behind a global database identifier.

Dataset

A Dataset is one query space rooted at a normalized local path or object-store URI. The path is its identity. A Warehouse mount name is only a SQL alias. A dataset pin (@name) or Directory library name is a locator; after resolution the engine opens the path.

A Dataset is a discovery, Snapshot, query, and exchange boundary. It does not claim that every expected external task produced a Source. pChronicle reports the Sources it can discover and pin; it does not infer unreported trajectories.

Source

A Source is the smallest independently discovered and versioned trajectory representation inside a Dataset. It may be a canonical event store, a Storyline projection, or a supported exchange file. Every normalized row keeps its source_path, so external IDs remain Source-local and collisions stay visible.

The complete address of an entity is therefore:

(path, source_path, entity_kind, original_id)

Snapshot

A Snapshot is the sync protocol between writers and readers after a path is opened. It fixes the Source membership and version references used by one operation so a query does not silently switch versions mid-scan. It does not claim that unrelated Sources were produced at one global instant.

Directory listing and tickets (RFC-0013) are not Snapshots. They only decide which paths a caller may open.

Use the Dataset workflow to inspect these objects. The implementation is described by the Snapshot design.