statistics: using standard lib slices instead of exp slices (#45996)

ref #45933

statistics/BUILD.bazel

@@ -121,7 +121,6 @@ go_test(
         "@com_github_pingcap_failpoint//:failpoint",
         "@com_github_pingcap_log//:log",
         "@com_github_stretchr_testify//require",
-        "@org_golang_x_exp//slices",
         "@org_uber_go_goleak//:goleak",
         "@org_uber_go_zap//:zap",
     ],

statistics/cmsketch.go

@@ -16,9 +16,11 @@ package statistics
 
 import (
 	"bytes"
+	"cmp"
 	"fmt"
 	"math"
 	"reflect"
+	"slices"
 	"sort"
 	"strings"
 	"sync/atomic"
@@ -39,7 +41,6 @@ import (
 	"github.com/pingcap/tidb/util/mathutil"
 	"github.com/pingcap/tipb/go-tipb"
 	"github.com/twmb/murmur3"
-	"golang.org/x/exp/slices"
 )
 
 // topNThreshold is the minimum ratio of the number of topN elements in CMSketch, 10 means 1 / 10 = 10%.
@@ -265,12 +266,12 @@ func queryValue(sctx sessionctx.Context, c *CMSketch, t *TopN, val types.Datum)
 	if sctx != nil {
 		sc = sctx.GetSessionVars().StmtCtx
 	}
-	bytes, err := tablecodec.EncodeValue(sc, nil, val)
+	rawData, err := tablecodec.EncodeValue(sc, nil, val)
 	if err != nil {
 		return 0, errors.Trace(err)
 	}
-	h1, h2 := murmur3.Sum128(bytes)
-	if ret, ok := t.QueryTopN(sctx, bytes); ok {
+	h1, h2 := murmur3.Sum128(rawData)
+	if ret, ok := t.QueryTopN(sctx, rawData); ok {
 		return ret, nil
 	}
 	return c.queryHashValue(sctx, h1, h2), nil
@@ -289,7 +290,7 @@ func (c *CMSketch) QueryBytes(d []byte) uint64 {
 func (c *CMSketch) queryHashValue(sctx sessionctx.Context, h1, h2 uint64) (result uint64) {
 	vals := make([]uint32, c.depth)
 	originVals := make([]uint32, c.depth)
-	min := uint32(math.MaxUint32)
+	minValue := uint32(math.MaxUint32)
 	useDefaultValue := false
 	if sctx != nil && sctx.GetSessionVars().StmtCtx.EnableOptimizerDebugTrace {
 		debugtrace.EnterContextCommon(sctx)
@@ -309,8 +310,8 @@ func (c *CMSketch) queryHashValue(sctx sessionctx.Context, h1, h2 uint64) (resul
 	for i := range c.table {
 		j := (h1 + h2*uint64(i)) % uint64(c.width)
 		originVals[i] = c.table[i][j]
-		if min > c.table[i][j] {
-			min = c.table[i][j]
+		if minValue > c.table[i][j] {
+			minValue = c.table[i][j]
 		}
 		noise := (c.count - uint64(c.table[i][j])) / (uint64(c.width) - 1)
 		if uint64(c.table[i][j]) == 0 {
@@ -323,8 +324,8 @@ func (c *CMSketch) queryHashValue(sctx sessionctx.Context, h1, h2 uint64) (resul
 	}
 	slices.Sort(vals)
 	res := vals[(c.depth-1)/2] + (vals[c.depth/2]-vals[(c.depth-1)/2])/2
-	if res > min+temp {
-		res = min + temp
+	if res > minValue+temp {
+		res = minValue + temp
 	}
 	if res == 0 {
 		return uint64(0)
@@ -641,11 +642,11 @@ func (c *TopN) findTopN(d []byte) int {
 	}
 	match := false
 	idx := sort.Search(len(c.TopN), func(i int) bool {
-		cmp := bytes.Compare(c.TopN[i].Encoded, d)
-		if cmp == 0 {
+		cmpRst := bytes.Compare(c.TopN[i].Encoded, d)
+		if cmpRst == 0 {
 			match = true
 		}
-		return cmp >= 0
+		return cmpRst >= 0
 	})
 	if !match {
 		return -1
@@ -660,11 +661,11 @@ func (c *TopN) LowerBound(d []byte) (idx int, match bool) {
 		return 0, false
 	}
 	idx = sort.Search(len(c.TopN), func(i int) bool {
-		cmp := bytes.Compare(c.TopN[i].Encoded, d)
-		if cmp == 0 {
+		cmpRst := bytes.Compare(c.TopN[i].Encoded, d)
+		if cmpRst == 0 {
 			match = true
 		}
-		return cmp >= 0
+		return cmpRst >= 0
 	})
 	return idx, match
 }
@@ -699,8 +700,8 @@ func (c *TopN) Sort() {
 	if c == nil {
 		return
 	}
-	slices.SortFunc(c.TopN, func(i, j TopNMeta) bool {
-		return bytes.Compare(i.Encoded, j.Encoded) < 0
+	slices.SortFunc(c.TopN, func(i, j TopNMeta) int {
+		return bytes.Compare(i.Encoded, j.Encoded)
 	})
 }
 
@@ -919,11 +920,11 @@ func checkEmptyTopNs(topNs []*TopN) bool {
 
 // SortTopnMeta sort topnMeta
 func SortTopnMeta(topnMetas []TopNMeta) []TopNMeta {
-	slices.SortFunc(topnMetas, func(i, j TopNMeta) bool {
+	slices.SortFunc(topnMetas, func(i, j TopNMeta) int {
 		if i.Count != j.Count {
-			return i.Count > j.Count
+			return cmp.Compare(j.Count, i.Count)
 		}
-		return bytes.Compare(i.Encoded, j.Encoded) < 0
+		return bytes.Compare(i.Encoded, j.Encoded)
 	})
 	return topnMetas
 }
@@ -934,11 +935,11 @@ func GetMergedTopNFromSortedSlice(sorted []TopNMeta, n uint32) (*TopN, []TopNMet
 }
 
 func getMergedTopNFromSortedSlice(sorted []TopNMeta, n uint32) (*TopN, []TopNMeta) {
-	slices.SortFunc(sorted, func(i, j TopNMeta) bool {
+	slices.SortFunc(sorted, func(i, j TopNMeta) int {
 		if i.Count != j.Count {
-			return i.Count > j.Count
+			return cmp.Compare(j.Count, i.Count)
 		}
-		return bytes.Compare(i.Encoded, j.Encoded) < 0
+		return bytes.Compare(i.Encoded, j.Encoded)
 	})
 	n = mathutil.Min(uint32(len(sorted)), n)
 

statistics/cmsketch_test.go

@@ -339,6 +339,18 @@ func TestMergePartTopN2GlobalTopNWithoutHists(t *testing.T) {
 	require.Len(t, leftTopN, 1, "should have 1 left topN")
 }
 
+func TestSortTopnMeta(t *testing.T) {
+	data := []TopNMeta{{
+		Encoded: []byte("a"),
+		Count:   1,
+	}, {
+		Encoded: []byte("b"),
+		Count:   2,
+	}}
+	sortedData := SortTopnMeta(data)
+	require.Equal(t, uint64(2), sortedData[0].Count)
+}
+
 func TestMergePartTopN2GlobalTopNWithHists(t *testing.T) {
 	loc := time.UTC
 	sc := &stmtctx.StatementContext{TimeZone: loc}

statistics/feedback.go

@@ -19,6 +19,7 @@ import (
 	"encoding/gob"
 	"math"
 	"math/rand"
+	"slices"
 	"sort"
 	goatomic "sync/atomic"
 	"time"
@@ -41,7 +42,6 @@ import (
 	"github.com/pingcap/tidb/util/ranger"
 	"go.uber.org/atomic"
 	"go.uber.org/zap"
-	"golang.org/x/exp/slices"
 )
 
 // Feedback represents the total scan count in range [lower, upper).
@@ -353,19 +353,19 @@ func NonOverlappedFeedbacks(sc *stmtctx.StatementContext, fbs []Feedback) ([]Fee
 	// Sort feedbacks by end point and start point incrementally, then pick every feedback that is not overlapped
 	// with the previous chosen feedbacks.
 	var existsErr bool
-	slices.SortFunc(fbs, func(i, j Feedback) bool {
+	slices.SortFunc(fbs, func(i, j Feedback) int {
 		res, err := i.Upper.Compare(sc, j.Upper, collate.GetBinaryCollator())
 		if err != nil {
 			existsErr = true
 		}
 		if existsErr || res != 0 {
-			return res < 0
+			return res
 		}
 		res, err = i.Lower.Compare(sc, j.Lower, collate.GetBinaryCollator())
 		if err != nil {
 			existsErr = true
 		}
-		return res < 0
+		return res
 	})
 	if existsErr {
 		return fbs, false

statistics/handle/BUILD.bazel

@@ -49,7 +49,6 @@ go_library(
         "@com_github_pingcap_log//:log",
         "@com_github_pingcap_tipb//go-tipb",
         "@com_github_tikv_client_go_v2//oracle",
-        "@org_golang_x_exp//slices",
         "@org_uber_go_atomic//:atomic",
         "@org_uber_go_zap//:zap",
     ],

statistics/handle/handle.go

@@ -19,6 +19,7 @@ import (
 	"context"
 	"encoding/json"
 	"fmt"
+	"slices"
 	"strconv"
 	"strings"
 	"sync"
@@ -50,7 +51,6 @@ import (
 	"github.com/tikv/client-go/v2/oracle"
 	atomic2 "go.uber.org/atomic"
 	"go.uber.org/zap"
-	"golang.org/x/exp/slices"
 )
 
 const (
@@ -976,9 +976,8 @@ func MergeGlobalStatsTopNByConcurrency(mergeConcurrency, mergeBatchSize int, wra
 			// Remove the value from the Hists.
 			if len(removeTopn) > 0 {
 				tmp := removeTopn
-				slices.SortFunc(tmp, func(i, j statistics.TopNMeta) bool {
-					cmpResult := bytes.Compare(i.Encoded, j.Encoded)
-					return cmpResult < 0
+				slices.SortFunc(tmp, func(i, j statistics.TopNMeta) int {
+					return bytes.Compare(i.Encoded, j.Encoded)
 				})
 				wrapper.AllHg[i].RemoveVals(tmp)
 			}

statistics/handle/update.go

@@ -16,10 +16,12 @@ package handle
 
 import (
 	"bytes"
+	"cmp"
 	"context"
 	"fmt"
 	"math"
 	"math/rand"
+	"slices"
 	"strconv"
 	"strings"
 	"sync"
@@ -48,7 +50,6 @@ import (
 	"github.com/pingcap/tidb/util/timeutil"
 	"go.uber.org/atomic"
 	"go.uber.org/zap"
-	"golang.org/x/exp/slices"
 )
 
 type tableDeltaMap map[int64]variable.TableDelta
@@ -961,11 +962,11 @@ func (h *Handle) DumpColStatsUsageToKV() error {
 	for id, t := range colMap {
 		pairs = append(pairs, pair{tblColID: id, lastUsedAt: t.UTC().Format(types.TimeFormat)})
 	}
-	slices.SortFunc(pairs, func(i, j pair) bool {
+	slices.SortFunc(pairs, func(i, j pair) int {
 		if i.tblColID.TableID == j.tblColID.TableID {
-			return i.tblColID.ID < j.tblColID.ID
+			return cmp.Compare(i.tblColID.ID, j.tblColID.ID)
 		}
-		return i.tblColID.TableID < j.tblColID.TableID
+		return cmp.Compare(i.tblColID.TableID, j.tblColID.TableID)
 	})
 	// Use batch insert to reduce cost.
 	for i := 0; i < len(pairs); i += batchInsertSize {

statistics/index.go

@@ -17,6 +17,7 @@ package statistics
 import (
 	"bytes"
 	"math"
+	"slices"
 	"strings"
 
 	"github.com/pingcap/failpoint"
@@ -33,7 +34,6 @@ import (
 	"github.com/pingcap/tidb/util/mathutil"
 	"github.com/pingcap/tidb/util/ranger"
 	"github.com/twmb/murmur3"
-	"golang.org/x/exp/slices"
 )
 
 // Index represents an index histogram.

statistics/selectivity_test.go

@@ -21,6 +21,7 @@ import (
 	"os"
 	"regexp"
 	"runtime/pprof"
+	"slices"
 	"testing"
 	"time"
 
@@ -44,7 +45,6 @@ import (
 	"github.com/pingcap/tidb/util/mock"
 	"github.com/pingcap/tidb/util/ranger"
 	"github.com/stretchr/testify/require"
-	"golang.org/x/exp/slices"
 )
 
 func TestCollationColumnEstimate(t *testing.T) {