fix

Author: forsaken628

src/query/sql/src/planner/plans/union_all.rs

@@ -133,6 +133,9 @@ impl UnionAll {
                         {
                             return Ok(Some((output, right)));
                         }
+                        // TODO: Distinguish all-NULL column statistics from unknown statistics so
+                        // a non-empty all-NULL branch can contribute its NULL count without
+                        // discarding the other branch's value statistics.
                         _ => return Ok(None),
                     };
                     let mut ndv = Self::merge_ndv(&left, &right)?;
@@ -179,6 +182,14 @@ impl UnionAll {
 
     fn merge_ndv(left: &ColumnStat, right: &ColumnStat) -> Result<NdvEstimate> {
         let ndv_upper = left.ndv.upper + right.ndv.upper;
+        let ranges_disjoint = left.max.compare(&right.min)? == Ordering::Less
+            || right.max.compare(&left.min)? == Ordering::Less;
+        if ranges_disjoint
+            && let (Some(left), Some(right)) = (left.ndv.expected, right.ndv.expected)
+        {
+            return Ok(NdvEstimate::new(left + right, ndv_upper));
+        }
+
         if left.min.is_numeric()
             && let (Some(left), Some(left_ndv)) = (&left.histogram, left.ndv.expected)
             && let (Some(right), Some(right_ndv)) = (&right.histogram, right.ndv.expected)

src/query/sql/tests/it/optimizer/union_all.rs

@@ -261,7 +261,7 @@ fn cases() -> Result<Vec<UnionCase>> {
     Ok(vec![
         UnionCase {
             name: "lossless_coercion",
-            description: "SQL binding inserts a nullable INT to BIGINT coercion and UnionAll derives statistics after the cast.",
+            description: "SQL binding inserts a nullable INT to BIGINT coercion; after the cast drops the histogram, disjoint ranges still make the union NDV additive.",
             sql: "SELECT k FROM l UNION ALL SELECT k FROM r",
             tables: vec![
                 table(

src/query/sql/tests/it/optimizer/union_all.txt

@@ -1,5 +1,5 @@
 === lossless_coercion ===
-description: SQL binding inserts a nullable INT to BIGINT coercion and UnionAll derives statistics after the cast.
+description: SQL binding inserts a nullable INT to BIGINT coercion; after the cast drops the histogram, disjoint ranges still make the union NDV additive.
 sql: SELECT k FROM l UNION ALL SELECT k FROM r
 path: []
 node:
@@ -22,7 +22,7 @@ UnionAll
 
 cardinality: 30.000
 precise_cardinality: Some(30)
-column_stat: k (#2) min=1, max=8, ndv=~5.0[..8.0], null_count=3, histogram=none
+column_stat: k (#2) min=1, max=8, ndv=8.0, null_count=3, histogram=none
 
 === finite_range_reduces_union_ndv_bound ===
 description: UnionAll reduces its conservative NDV upper bound to the merged finite range before an outer cast consumes the statistics.