* Enter bulk-edge-load mode: drop the non-unique edge indexes so the mass * INSERT OR IGNORE stream pays one B-tree (the identity index) instead of * five — measured 2.8s → 1.1s inserting a 224k-edge resolution set, with * recreation costing ~0.3s. MUST be paired with endBulkEdgeLoad(); a c
()
| 418 | * re-applies CREATE INDEX IF NOT EXISTS). |
| 419 | */ |
| 420 | beginBulkEdgeLoad(): void { |
| 421 | for (const idx of DatabaseConnection.BULK_EDGE_INDEX_NAMES) { |
| 422 | this.db.exec(`DROP INDEX IF EXISTS ${idx}`); |
| 423 | } |
| 424 | } |
| 425 | |
| 426 | /** |
| 427 | * Leave bulk-edge-load mode: recreate the dropped indexes in one pass each |