From 3e4cb34bfde181c45d1daf2098f4fa6f232e7c36 Mon Sep 17 00:00:00 2001
From: Pascal Seitz <pascal.seitz@gmail.com>
Date: Thu, 23 Jul 2026 12:27:21 +0200
Subject: [PATCH] add missing to benchmark

---
 benches/agg_bench.rs                          | 43 +++++++++++++++++++
 .../src/column_index/optional_index/mod.rs    |  1 +
 2 files changed, 44 insertions(+)

diff --git a/benches/agg_bench.rs b/benches/agg_bench.rs
index 114774200..2c83461c8 100644
--- a/benches/agg_bench.rs
+++ b/benches/agg_bench.rs
@@ -87,6 +87,7 @@ fn bench_agg(mut group: InputGroup<Index>) {
     register!(group, terms_many_with_avg_sub_agg);
     register!(group, terms_status_with_avg_sub_agg);
     register!(group, nested_terms_status_and_zipf_1000);
+    register!(group, nested_terms_status_and_zipf_1000_with_missing);
     register!(group, nested_terms_zipf_1000_and_status);
     register!(group, nested_terms_many_and_zipf_1000);
     register!(group, nested_terms_many_and_zipf_1000_and_status);
@@ -115,6 +116,11 @@ fn bench_agg(mut group: InputGroup<Index>) {
         multi_terms_status_and_zipf_1000,
         multi_terms_status_and_zipf_1000_filtered
     );
+    register!(
+        group,
+        multi_terms_status_and_zipf_1000_with_missing,
+        multi_terms_status_and_zipf_1000_with_missing_filtered
+    );
     register!(
         group,
         multi_terms_zipf_1000_and_status,
@@ -431,6 +437,26 @@ fn nested_terms_status_and_zipf_1000(index: &Index) {
     execute_agg(index, agg_req);
 }
 
+fn nested_terms_status_and_zipf_1000_with_missing(index: &Index) {
+    let agg_req = json!({
+        "my_texts": {
+            "terms": {
+                "field": "text_few_terms_status",
+                "missing": "MISSING_STATUS"
+            },
+            "aggs": {
+                "nested_terms": {
+                    "terms": {
+                        "field": "text_1000_terms_zipf",
+                        "missing": "MISSING_ZIPF"
+                    }
+                }
+            }
+        }
+    });
+    execute_agg(index, agg_req);
+}
+
 fn nested_terms_zipf_1000_and_status(index: &Index) {
     let agg_req = json!({
         "my_texts": {
@@ -717,6 +743,23 @@ define_multi_terms_benchmark!(
     }),
 );
 
+define_multi_terms_benchmark!(
+    /// multi_terms equivalent of nested_terms_status_and_zipf_1000_with_missing.
+    multi_terms_status_and_zipf_1000_with_missing,
+    multi_terms_status_and_zipf_1000_with_missing_filtered,
+    json!({
+        "mt": {
+            "multi_terms": {
+                "terms": [
+                    {"field": "text_few_terms_status", "missing": "MISSING_STATUS"},
+                    {"field": "text_1000_terms_zipf", "missing": "MISSING_ZIPF"}
+                ],
+                "size": 100
+            }
+        }
+    }),
+);
+
 define_multi_terms_benchmark!(
     /// multi_terms equivalent of nested_terms_zipf_1000_and_status:
     /// flat GroupBy(zipf_1000, status) vs nested terms(zipf_1000) -> terms(status)
diff --git a/columnar/src/column_index/optional_index/mod.rs b/columnar/src/column_index/optional_index/mod.rs
index 8dc7086d4..dd55994fe 100644
--- a/columnar/src/column_index/optional_index/mod.rs
+++ b/columnar/src/column_index/optional_index/mod.rs
@@ -291,6 +291,7 @@ impl OptionalIndex {
         doc_ids_out: &mut Vec<DocId>,
         row_ids_out: &mut Vec<RowId>,
     ) {
+        // TODO: change the API to require sorted input, and remove the fallback.
         if !doc_ids.is_sorted() {
             for &doc_id in doc_ids {
                 if let Some(row_id) = self.rank_if_exists(doc_id) {