Merge pull request #155651 from michae2/backport25.2-155035

release-25.2: sql/stats: handle range counts when skipping dropped enum hist buckets

Author: michae2

pkg/sql/opt/cat/table.go

@@ -295,7 +295,7 @@ type HistogramBucket struct {
 
 	// DistinctRange is the estimated number of distinct values between the upper
 	// bound of the previous bucket and UpperBound (both boundaries are
-	// exclusive).
+	// exclusive). The first bucket should always have DistinctRange=0.
 	DistinctRange float64
 
 	// UpperBound is the upper bound of the bucket.

pkg/sql/opt/props/histogram.go

@@ -23,6 +23,7 @@ import (
 	"github.com/cockroachdb/cockroach/pkg/sql/sem/eval"
 	"github.com/cockroachdb/cockroach/pkg/sql/sem/tree"
 	"github.com/cockroachdb/cockroach/pkg/sql/types"
+	"github.com/cockroachdb/cockroach/pkg/util/buildutil"
 	"github.com/cockroachdb/cockroach/pkg/util/encoding"
 	"github.com/cockroachdb/cockroach/pkg/util/timetz"
 	"github.com/cockroachdb/errors"
@@ -79,6 +80,15 @@ func (h *Histogram) numEq(i int) float64 {
 // selectivity applied. i must be greater than or equal to 0 and less than
 // bucketCount.
 func (h *Histogram) numRange(i int) float64 {
+	// The first bucket always has a zero value for NumRange, so the lower bound
+	// of the histogram is the upper bound of the first bucket. We only check this
+	// in test builds.
+	if i == 0 && h.buckets[i].NumRange != 0 {
+		if buildutil.CrdbTestBuild {
+			panic(errors.AssertionFailedf("the first bucket should have NumRange=0"))
+		}
+		return 0
+	}
 	return h.buckets[i].NumRange * h.selectivity
 }
 
@@ -89,6 +99,13 @@ func (h *Histogram) distinctRange(i int) float64 {
 	n := h.buckets[i].NumRange
 	d := h.buckets[i].DistinctRange
 
+	if i == 0 && d != 0 {
+		if buildutil.CrdbTestBuild {
+			panic(errors.AssertionFailedf("the first bucket should have DistinctRange=0"))
+		}
+		return 0
+	}
+
 	if d == 0 {
 		return 0
 	}
@@ -207,11 +224,6 @@ func (h *Histogram) maxDistinctValuesCount() float64 {
 		return 0
 	}
 
-	// The first bucket always has a zero value for NumRange, so the lower bound
-	// of the histogram is the upper bound of the first bucket.
-	if h.numRange(0) != 0 {
-		panic(errors.AssertionFailedf("the first bucket should have NumRange=0"))
-	}
 	previousUpperBound := h.upperBound(0)
 
 	var count float64
@@ -317,12 +329,6 @@ func (h *Histogram) filter(
 		return filtered
 	}
 
-	// The first bucket always has a zero value for NumRange, so the lower bound
-	// of the histogram is the upper bound of the first bucket.
-	if h.numRange(0) != 0 {
-		panic(errors.AssertionFailedf("the first bucket should have NumRange=0"))
-	}
-
 	var iter histogramIter
 	iter.init(h, desc)
 	spanIndex := 0