Merge #144127

144127: workload/tpcc: temporarily add back golang.org/x/exp/rand r=yuzefovich a=yuzefovich

This commit is partial revert of 84dd73a which removed the usages of golang.org/x/exp/rand from the code base. In particular, this commit brings back the usage into Order and OrderLine generation of TPCC fixtures since the mentioned commit broke the mixed-version import. It's not immediately clear what's wrong there, but the imported tables in the mixed-version 25.1-master state don't pass 3.3.2.4 check.

I've been using COCKROACH_RANDOM_SEED=-1621471749504012229 with `import/mixed-versions` for (relatively) fast reproduction.

Informs: #143870.
Epic: None
Release note: None

Co-authored-by: Yahor Yuzefovich <[email protected]>

Author: craig[bot]

pkg/ccl/workloadccl/allccl/all_test.go

@@ -282,7 +282,7 @@ func TestDeterministicInitialData(t *testing.T) {
 		`roachmart`:  0xda5e73423dbdb2d9,
 		`sqlsmith`:   0xcbf29ce484222325,
 		`startrek`:   0xa0249fbdf612734c,
-		`tpcc`:       0x3f37ea71beae16fb,
+		`tpcc`:       0xcccced25deea244e,
 		`tpch`:       0xcd2abbd021ed895d,
 		`ycsb`:       0x0e6012ee6491a0fb,
 	}

pkg/testutils/lint/lint_test.go

@@ -2090,6 +2090,8 @@ func TestLint(t *testing.T) {
 			stream.GrepNot(`pkg/cmd/mirror/go/mirror.go`),
 			// As above, the bazel build tag has an impact here.
 			stream.GrepNot(`pkg/testutils/docker/single_node_docker_test.go`),
+			// TODO(#143870): remove uses of this package.
+			stream.GrepNot(`"golang.org/x/exp/rand" is deprecated`),
 		}
 		for analyzerName, config := range nogoConfig {
 			if !staticcheckCheckNameRe.MatchString(analyzerName) {

pkg/workload/tpcc/BUILD.bazel

@@ -50,6 +50,7 @@ go_library(
         "@com_github_prometheus_client_golang//prometheus",
         "@com_github_prometheus_client_golang//prometheus/promauto",
         "@com_github_spf13_pflag//:pflag",
+        "@org_golang_x_exp//rand",
         "@org_golang_x_sync//errgroup",
     ],
 )

pkg/workload/tpcc/generate.go

@@ -14,6 +14,7 @@ import (
 	"github.com/cockroachdb/cockroach/pkg/util/bufalloc"
 	"github.com/cockroachdb/cockroach/pkg/util/uuid"
 	"github.com/cockroachdb/cockroach/pkg/workload"
+	randold "golang.org/x/exp/rand"
 )
 
 // These constants are all set by the spec - they're not knobs. Don't change
@@ -57,6 +58,7 @@ const (
 
 type generateLocals struct {
 	rng       tpccRand
+	rngOld    tpccRandOld
 	uuidAlloc uuid.UUID
 }
 
@@ -480,8 +482,8 @@ func (w *tpcc) tpccOrderInitialRowBatch(rowIdx int, cb coldata.Batch, a *bufallo
 
 	// NB: numOrderLines is not allowed to use precomputed random data, make sure
 	// it stays that way. See 4.3.2.1.
-	l.rng.Rand = rand.New(rand.NewPCG(RandomSeed.Seed(), uint64(rowIdx)))
-	numOrderLines := randInt(l.rng.Rand, minOrderLinesPerOrder, maxOrderLinesPerOrder)
+	l.rngOld.Seed(RandomSeed.Seed() + uint64(rowIdx))
+	numOrderLines := randIntOld(l.rngOld.Rand, minOrderLinesPerOrder, maxOrderLinesPerOrder)
 
 	oID := (rowIdx % numOrdersPerDistrict) + 1
 	dID := ((rowIdx / numOrdersPerDistrict) % numDistrictsPerWarehouse) + 1
@@ -500,7 +502,7 @@ func (w *tpcc) tpccOrderInitialRowBatch(rowIdx int, cb coldata.Batch, a *bufallo
 			// We need a random permutation of customers that stable for all orders in a
 			// district, so use the district ID to seed the random permutation.
 			w.randomCIDsCache.values[dID] = make([]int, numCustomersPerDistrict)
-			for i, cID := range rand.New(rand.NewPCG(uint64(dID), 0)).Perm(numCustomersPerDistrict) {
+			for i, cID := range randold.New(randold.NewSource(uint64(dID))).Perm(numCustomersPerDistrict) {
 				w.randomCIDsCache.values[dID][i] = cID + 1
 			}
 		}
@@ -512,7 +514,7 @@ func (w *tpcc) tpccOrderInitialRowBatch(rowIdx int, cb coldata.Batch, a *bufallo
 	var carrierID int64
 	if oID < 2101 {
 		carrierSet = true
-		carrierID = randInt(l.rng.Rand, 1, 10)
+		carrierID = randIntOld(l.rngOld.Rand, 1, 10)
 	}
 
 	cb.Reset(orderTypes, 1, coldata.StandardColumnFactory)
@@ -601,15 +603,15 @@ func (w *tpcc) tpccOrderLineInitialRowBatch(
 
 	// NB: numOrderLines is not allowed to use precomputed random data, make sure
 	// it stays that way. See 4.3.2.1.
-	l.rng.Rand = rand.New(rand.NewPCG(RandomSeed.Seed(), uint64(orderRowIdx)))
-	numOrderLines := int(randInt(l.rng.Rand, minOrderLinesPerOrder, maxOrderLinesPerOrder))
+	l.rngOld.Seed(RandomSeed.Seed() + uint64(orderRowIdx))
+	numOrderLines := int(randIntOld(l.rngOld.Rand, minOrderLinesPerOrder, maxOrderLinesPerOrder))
 
 	// NB: There is one batch of order_line rows per order
 	oID := (orderRowIdx % numOrdersPerDistrict) + 1
 	dID := ((orderRowIdx / numOrdersPerDistrict) % numDistrictsPerWarehouse) + 1
 	wID := (orderRowIdx / numOrdersPerWarehouse)
 
-	ao := aCharsOffset(l.rng.IntN(len(aCharsAlphabet)))
+	ao := aCharsOffset(l.rngOld.Intn(len(aCharsAlphabet)))
 	cb.Reset(orderLineTypes, numOrderLines, coldata.StandardColumnFactory)
 	olOIDCol := cb.ColVec(0).Int64()
 	olDIDCol := cb.ColVec(1).Int64()
@@ -634,14 +636,14 @@ func (w *tpcc) tpccOrderLineInitialRowBatch(
 			amount = 0
 			deliveryDSet = true
 		} else {
-			amount = float64(randInt(l.rng.Rand, 1, 999999)) / 100.0
+			amount = float64(randIntOld(l.rngOld.Rand, 1, 999999)) / 100.0
 		}
 
 		olOIDCol[rowIdx] = int64(oID)
 		olDIDCol[rowIdx] = int64(dID)
 		olWIDCol[rowIdx] = int64(wID)
 		olNumberCol[rowIdx] = int64(olNumber)
-		olIIDCol[rowIdx] = randInt(l.rng.Rand, 1, 100000)
+		olIIDCol[rowIdx] = randIntOld(l.rngOld.Rand, 1, 100000)
 		olSupplyWIDCol[rowIdx] = int64(wID)
 		if deliveryDSet {
 			olDeliveryDCol.Set(rowIdx, w.nowTime)
@@ -650,7 +652,7 @@ func (w *tpcc) tpccOrderLineInitialRowBatch(
 		}
 		olQuantityCol[rowIdx] = 5
 		olAmountCol[rowIdx] = amount
-		olDistInfoCol.Set(rowIdx, randAStringInitialDataOnly(&l.rng, &ao, a, 24, 24))
+		olDistInfoCol.Set(rowIdx, randAStringInitialDataOnlyOld(&l.rngOld, &ao, a, 24, 24))
 	}
 }
 

pkg/workload/tpcc/random.go

@@ -10,6 +10,7 @@ import (
 
 	"github.com/cockroachdb/cockroach/pkg/util/bufalloc"
 	"github.com/cockroachdb/cockroach/pkg/workload/workloadimpl"
+	randold "golang.org/x/exp/rand"
 )
 
 var cLastTokens = [...]string{
@@ -34,6 +35,13 @@ type tpccRand struct {
 	aChars, letters, numbers workloadimpl.PrecomputedRand
 }
 
+// TODO(#143870): remove this.
+type tpccRandOld struct {
+	*randold.Rand
+
+	aChars, letters, numbers workloadimpl.PrecomputedRand
+}
+
 type aCharsOffset int
 type lettersOffset int
 type numbersOffset int
@@ -59,6 +67,27 @@ func randStringFromAlphabet(
 	return b
 }
 
+func randStringFromAlphabetOld(
+	rng *randold.Rand,
+	a *bufalloc.ByteAllocator,
+	minLen, maxLen int,
+	pr workloadimpl.PrecomputedRand,
+	prOffset *int,
+) []byte {
+	size := maxLen
+	if maxLen-minLen != 0 {
+		size = int(randIntOld(rng, minLen, maxLen))
+	}
+	if size == 0 {
+		return nil
+	}
+
+	var b []byte
+	*a, b = a.Alloc(size, 0 /* extraCap */)
+	*prOffset = pr.FillBytes(*prOffset, b)
+	return b
+}
+
 // randAStringInitialDataOnly generates a random alphanumeric string of length
 // between min and max inclusive. It uses a set of pregenerated random data,
 // which the spec allows only for initial data. See 4.3.2.2.
@@ -71,6 +100,12 @@ func randAStringInitialDataOnly(
 	return randStringFromAlphabet(rng.Rand, a, min, max, rng.aChars, (*int)(ao))
 }
 
+func randAStringInitialDataOnlyOld(
+	rng *tpccRandOld, ao *aCharsOffset, a *bufalloc.ByteAllocator, min, max int,
+) []byte {
+	return randStringFromAlphabetOld(rng.Rand, a, min, max, rng.aChars, (*int)(ao))
+}
+
 // randNStringInitialDataOnly generates a random numeric string of length
 // between min and max inclusive. See 4.3.2.2.
 //
@@ -138,6 +173,10 @@ func randInt(rng *rand.Rand, min, max int) int64 {
 	return int64(rng.IntN(max-min+1) + min)
 }
 
+func randIntOld(rng *randold.Rand, min, max int) int64 {
+	return int64(rng.Intn(max-min+1) + min)
+}
+
 // randCLastSyllables returns a customer last name string generated according to
 // the table in 4.3.2.3. Given a number between 0 and 999, each of the three
 // syllables is determined by the corresponding digit in the three digit

pkg/workload/tpcc/tpcc.go

@@ -31,6 +31,7 @@ import (
 	"github.com/cockroachdb/errors"
 	"github.com/jackc/pgx/v5"
 	"github.com/spf13/pflag"
+	randold "golang.org/x/exp/rand"
 	"golang.org/x/sync/errgroup"
 )
 
@@ -661,6 +662,9 @@ func (w *tpcc) Tables() []workload.Table {
 	aCharsInit := workloadimpl.PrecomputedRandInit(rand.NewPCG(seed, 0), precomputedLength, aCharsAlphabet)
 	lettersInit := workloadimpl.PrecomputedRandInit(rand.NewPCG(seed, 0), precomputedLength, lettersAlphabet)
 	numbersInit := workloadimpl.PrecomputedRandInit(rand.NewPCG(seed, 0), precomputedLength, numbersAlphabet)
+	aCharsInitOld := workloadimpl.PrecomputedRandInit(randold.New(randold.NewSource(seed)), precomputedLength, aCharsAlphabet)
+	lettersInitOld := workloadimpl.PrecomputedRandInit(randold.New(randold.NewSource(seed)), precomputedLength, lettersAlphabet)
+	numbersInitOld := workloadimpl.PrecomputedRandInit(randold.New(randold.NewSource(seed)), precomputedLength, numbersAlphabet)
 	if w.localsPool == nil {
 		w.localsPool = &sync.Pool{
 			New: func() interface{} {
@@ -674,6 +678,15 @@ func (w *tpcc) Tables() []workload.Table {
 						letters: lettersInit(),
 						numbers: numbersInit(),
 					},
+					rngOld: tpccRandOld{
+						Rand: randold.New(randold.NewSource(uint64(timeutil.Now().UnixNano()))),
+						// Intentionally wait until here to initialize the precomputed rands
+						// so a caller of Tables that only wants schema doesn't compute
+						// them.
+						aChars:  aCharsInitOld(),
+						letters: lettersInitOld(),
+						numbers: numbersInitOld(),
+					},
 				}
 			},
 		}