@inerrata-corporation/errata 2.0.23-dev.4734 → 2.0.23-dev.4796

package/consolidate-worker.mjs

@@ -17613,7 +17613,8 @@ var SELECT_INDEXES = [
   "nodes_live_fold_candidate",
   "nodes_live_resolved",
   "nodes_live_fix_candidate",
-  "nodes_live_anchor_hint"
+  "nodes_live_anchor_hint",
+  "nodes_live_created_v2"
 ];
 var EMBEDDING_DIM_SQL = "(CASE WHEN length(embedding_blob) >= 4 THEN length(embedding_blob) / 4 ELSE COALESCE(json_array_length(embedding_json), 0) END)";
 var KEY = /^[A-Za-z_][A-Za-z0-9_]*(\.[A-Za-z_][A-Za-z0-9_]*)*$/;
@@ -17934,7 +17935,10 @@ CREATE UNIQUE INDEX IF NOT EXISTS edges_triple ON edges(from_id, type, to_id);
 `;
 var TEMPORAL_INDEX_SQL = `
 CREATE INDEX IF NOT EXISTS nodes_live ON nodes(label, valid_to);
-CREATE INDEX IF NOT EXISTS nodes_live_count ON nodes(valid_to);
+-- Dropped: a prefix of nodes_live_page_rank (valid_to, page_rank), which serves
+-- the live count and the "any live node?" check as a covering index too. It was
+-- re-keyed on every close and revive: ~11 KB of a symbol flip's ~80 KB.
+DROP INDEX IF EXISTS nodes_live_count;
 -- Problems by recency (the only label read newest-first). Over every live row,
 -- each write that stamps last_updated_at moved a code symbol's entry from a
 -- scattered leaf: ~19 KB per position-only symbol rewrite on a 1.1 GB store.
@@ -17946,7 +17950,13 @@ CREATE INDEX IF NOT EXISTS nodes_live_problem_recent ON nodes(label, last_update
 -- walks nodes_live and reads every live row of the label to test created_at:
 -- 56 MB per guard on a 1.1 GB store, to learn there was nothing to do. valid_to
 -- is a key column so the planner prefers this over nodes_live (one more term).
-CREATE INDEX IF NOT EXISTS nodes_live_created ON nodes(label, valid_to, created_at) WHERE valid_to IS NULL;
+-- last_updated_at rides along for errata.recent on code labels, which names this
+-- index: the filter runs on index entries, not on every row of the label (79 ms
+-- \u2192 3 ms for Function on a 1.1 GB store). It costs no extra writes: a close or
+-- revive already re-keys this index, and mergeNodeIfChanged leaves a symbol's
+-- last_updated_at alone when only its position moved.
+DROP INDEX IF EXISTS nodes_live_created;
+CREATE INDEX IF NOT EXISTS nodes_live_created_v2 ON nodes(label, valid_to, created_at, last_updated_at) WHERE valid_to IS NULL;
 -- Community members (the skills pull reads them twice per session boundary):
 -- ~15% of semantic nodes carry a localCommunity, and without this every live
 -- semantic row was read to find them (168 MB \u2192 18 MB per read).
@@ -18131,6 +18141,12 @@ function sameSqlValue(a, b) {
   if (a instanceof Uint8Array && b instanceof Uint8Array) return Buffer.compare(a, b) === 0;
   return false;
 }
+var POSITION_ATTRS = ["startByte", "endByte", "startRow"];
+function attrsJsonWithoutPositions(attrs) {
+  const rest = { ...attrs };
+  for (const k of POSITION_ATTRS) delete rest[k];
+  return JSON.stringify(rest);
+}
 var PRIOR_TAG_MARKER = '"priorTag":true';
 var SqliteGraphStore = class {
   db;
@@ -18231,7 +18247,7 @@ var SqliteGraphStore = class {
       lastUpdatedOf: this.db.prepare("SELECT last_updated_at AS t, attrs_json FROM nodes WHERE id = ?"),
       labelOf: this.db.prepare("SELECT label FROM nodes WHERE id = ?"),
       nodeContent: this.db.prepare(
-        "SELECT label, description, extraction_confidence, extraction_source, stability, valid_to, attrs_json FROM nodes WHERE id = ?"
+        "SELECT label, description, extraction_confidence, extraction_source, stability, valid_to, last_updated_at, attrs_json FROM nodes WHERE id = ?"
       ),
       edgeContent: this.db.prepare(
         "SELECT confidence, valid_to, attrs_json FROM edges WHERE from_id = ? AND type = ? AND to_id = ?"
@@ -18644,11 +18660,23 @@ var SqliteGraphStore = class {
    *  rank, community, landmark, surprise, hits, tier and the vector belong to
    *  the passes that compute them, and a skipped write leaves them as they are
    *  where a merge would have reset them to the producer's defaults. Returns
-   *  whether it wrote. */
+   *  whether it wrote.
+   *
+   *  A live row whose only change is its position keeps its lastUpdatedAt, so
+   *  that stamp means "this symbol's content changed": what errata.recent lists
+   *  and the fix-candidate ask reads, and what lets nodes_live_created_v2 carry
+   *  it without being re-keyed by every edit above a symbol. A closed row is
+   *  always stamped: reviveReobserved reopens what this pass stamped. */
   mergeNodeIfChanged(node) {
     const row = this.stmts.nodeContent.get(node.id);
-    if (row && row.valid_to === null && row.label === node.label && row.description === node.description && row.extraction_confidence === node.extractionConfidence && row.stability === node.stability && !EXPOSURE_ATTRS.some((k) => node.attrs?.[k] !== void 0) && row.attrs_json === JSON.stringify(node.attrs ?? {})) return false;
-    this.mergeNode(node);
+    if (!row || row.valid_to !== null || row.label !== node.label || row.description !== node.description || row.extraction_confidence !== node.extractionConfidence || row.stability !== node.stability || EXPOSURE_ATTRS.some((k) => node.attrs?.[k] !== void 0)) {
+      this.mergeNode(node);
+      return true;
+    }
+    const attrs = node.attrs ?? {};
+    if (row.attrs_json === JSON.stringify(attrs)) return false;
+    const moved = attrsJsonWithoutPositions(JSON.parse(row.attrs_json)) === attrsJsonWithoutPositions(attrs);
+    this.mergeNode(moved ? { ...node, lastUpdatedAt: row.last_updated_at } : node);
     return true;
   }
   updateNode(id, patch) {
@@ -18728,8 +18756,8 @@ var SqliteGraphStore = class {
       stmt2 = this.db.prepare(
         // Ordered, not left to the planner: readers take "first row wins" (the
         // symbol lexicon's name winner and doc pick), and the order otherwise
-        // followed whichever partial index the planner chose. nodes_live_created
-        // serves this order without a sort.
+        // followed whichever partial index the planner chose. nodes_live_created_v2
+        // serves it, sorting rowids only within an equal created_at.
         `SELECT id, description, community, extraction_confidence${cols} FROM nodes WHERE label = ? AND valid_to IS NULL ORDER BY created_at, rowid`
       );
       this.fieldStmts.set(cacheKey, stmt2);