lora-database 0.17.0

LoraDB — embeddable in-memory graph database with Cypher query support.
Documentation
{
  "_meta": {
    "default_tolerance_pct": 10,
    "last_updated": "2026-09-29",
    "notes": "Captured from `cargo bench -p lora-database --bench memory -- --quick` on Apple Silicon. Numbers are bytes from `MemoryReport::total_bytes()` after the build phase (or after_query for chain/star/social which run a representative MATCH after building). Tolerance is per-scenario; a regression beyond `total * (1 + tolerance_pct/100)` fails the gate. Bump the baseline whenever an intentional storage-shape change shifts a scenario. 2026-09-29: property maps became key-sorted vectors, index postings inline single ids, adjacency lists inline two edges. These are MemoryReport estimates; `cargo run --release -p lora-database --example heap_probe` measures real allocations. Later the same day: slabs, adjacency and label indexes became chunked copy-on-write vectors and index maps copy-on-write shards (structural sharing for staged writes)."
  },
  "benchmarks": {
    "node_only_10k": {
      "total_bytes": 3668174,
      "bytes_per_node": 309.2,
      "tolerance_pct": 10
    },
    "chain_1000.build": {
      "total_bytes": 514627,
      "bytes_per_node": 173.3,
      "bytes_per_rel": 100.3,
      "tolerance_pct": 10
    },
    "chain_10000.build": {
      "total_bytes": 5161131,
      "bytes_per_node": 173.3,
      "bytes_per_rel": 101.2,
      "tolerance_pct": 10
    },
    "chain_100000.build": {
      "total_bytes": 51531835,
      "bytes_per_node": 173.1,
      "bytes_per_rel": 100.9,
      "tolerance_pct": 10
    },
    "star_1000.build": {
      "total_bytes": 346187,
      "bytes_per_node": 172.3,
      "bytes_per_rel": 99.3,
      "tolerance_pct": 10
    },
    "star_10000.build": {
      "total_bytes": 3594019,
      "bytes_per_node": 173.2,
      "bytes_per_rel": 100.2,
      "tolerance_pct": 10
    },
    "star_100000.build": {
      "total_bytes": 35465315,
      "bytes_per_node": 173.0,
      "bytes_per_rel": 99.9,
      "tolerance_pct": 10
    },
    "social_20k_x8.build": {
      "total_bytes": 42867985,
      "bytes_per_node": 383.9,
      "bytes_per_rel": 165.9,
      "tolerance_pct": 10
    },
    "vector_flat_5k_d128": {
      "total_bytes": 6615020,
      "bytes_per_node": 681.3,
      "tolerance_pct": 10
    },
    "vector_hnsw_5k_d128": {
      "total_bytes": 10296500,
      "bytes_per_node": 681.3,
      "tolerance_pct": 15
    }
  }
}