Merge #50784 #50825 #50881 #50894

50784: opt: add ScanPrivate.IsUnfiltered() helper function r=RaduBerinde a=mgartner

This commit adds a new method to `ScanPrivate`, `IsUnfiltered`, which
returns true if the scan is guaranteed to produce all rows in the table.
It returns false if it may not return all rows because it has a limit,
it is constrained, or it scans a partial index.

This commit updates a few lines of code that were manually performing
these checks.

Related context: #49933 (comment)

Release note: None

50825: geogfn: apply bounding box calculations for DWithin r=sumeerbhola a=otan

Expand a BoundingRect by a given radius (using s2.Cap) and apply
intersection to do fairly cheap bounding box calculations before doing
the expensive DWithin logic.

Release note: None

50881: Makefile: change GEOS symlink to a copy r=petermattis a=otan

Instead of symlinking to the correct GEOS directory, force a copy
following the symlinks instead. This dramatically simplifies release
scripts from needing to follow symlinks to add the file to the
archive/Docker container.

Release note: None

50894: mutations: remove session variable setting from postgres mutator r=mjibson a=rohany

Fixes #48087.

This PR ensures that SQLSmith doesn't send session variable setting
statements to Postgres in various comparison tests.

Release note: None

Co-authored-by: Marcus Gartner <[email protected]>
Co-authored-by: Oliver Tan <[email protected]>
Co-authored-by: Rohan Yadav <[email protected]>

Makefile

@@ -771,7 +771,8 @@ $(LIBSNAPPY): $(SNAPPY_DIR)/Makefile bin/uptodate .ALWAYS_REBUILD
 $(LIBGEOS): $(GEOS_DIR)/Makefile bin/uptodate .ALWAYS_REBUILD
 	@uptodate $@ $(GEOS_SRC_DIR) || $(MAKE) --no-print-directory -C $(GEOS_DIR) geos_c
 	mkdir -p $(DYN_LIB_DIR)
-	ln -sf $(GEOS_DIR)/lib/lib{geos,geos_c}.$(DYN_EXT) $(DYN_LIB_DIR)
+	rm -f $(DYN_LIB_DIR)/lib{geos,geos_c}.$(DYN_EXT)
+	cp -L $(GEOS_DIR)/$(if $(target-is-windows),bin,lib)/lib{geos,geos_c}.$(DYN_EXT) $(DYN_LIB_DIR)
 
 $(LIBPROJ): $(PROJ_DIR)/Makefile bin/uptodate .ALWAYS_REBUILD
 	@uptodate $@ $(PROJ_SRC_DIR) || $(MAKE) --no-print-directory -C $(PROJ_DIR) proj

pkg/geo/geo.go

@@ -472,7 +472,7 @@ func (g *Geography) AsS2(emptyBehavior EmptyBehavior) ([]s2.Region, error) {
 	return S2RegionsFromGeom(geomRepr, emptyBehavior)
 }
 
-// BoundingRect returns the bounding rectangle of the given Geography.
+// BoundingRect returns the bounding s2.Rect of the given Geography.
 func (g *Geography) BoundingRect() s2.Rect {
 	bbox := g.spatialObject.BoundingBox
 	if bbox == nil {
@@ -484,6 +484,11 @@ func (g *Geography) BoundingRect() s2.Rect {
 	}
 }
 
+// BoundingCap returns the bounding s2.Cap of the given Geography.
+func (g *Geography) BoundingCap() s2.Cap {
+	return g.BoundingRect().CapBound()
+}
+
 // IsLinearRingCCW returns whether a given linear ring is counter clock wise.
 // See 2.07 of http://www.faqs.org/faqs/graphics/algorithms-faq/.
 // "Find the lowest vertex (or, if  there is more than one vertex with the same lowest coordinate,

pkg/geo/geogfn/distance.go

@@ -21,6 +21,12 @@ import (
 	"github.com/golang/geo/s2"
 )
 
+// SpheroidErrorFraction is an error fraction to compensate for using a sphere
+// to calculate the distance for what is actually a spheroid. The distance
+// calculation has an error that is bounded by (2 * spheroid.Flattening)%.
+// This 5% margin is pretty safe.
+const SpheroidErrorFraction = 0.05
+
 // Distance returns the distance between geographies a and b on a sphere or spheroid.
 // Returns a geo.EmptyGeometryError if any of the Geographies are EMPTY.
 func Distance(
@@ -217,12 +223,6 @@ func distanceGeographyRegions(
 	return minDistance, nil
 }
 
-// SpheroidErrorFraction is an error fraction to compensate for using a sphere
-// to calculate the distance for what is actually a spheroid. The distance
-// calculation has an error that is bounded by (2 * spheroid.Flattening)%.
-// This 5% margin is pretty safe.
-const SpheroidErrorFraction = 0.05
-
 // geographyMinDistanceUpdater finds the minimum distance using a sphere.
 // Methods will return early if it finds a minimum distance <= stopAfterLE.
 type geographyMinDistanceUpdater struct {

pkg/geo/geogfn/dwithin.go

@@ -13,6 +13,7 @@ package geogfn
 import (
 	"github.com/cockroachdb/cockroach/pkg/geo"
 	"github.com/cockroachdb/errors"
+	"github.com/golang/geo/s1"
 )
 
 // DWithin returns whether a is within distance d of b, i.e. Distance(a, b) <= d.
@@ -26,6 +27,18 @@ func DWithin(
 	if distance < 0 {
 		return false, errors.Newf("dwithin distance cannot be less than zero")
 	}
+	spheroid, err := a.Spheroid()
+	if err != nil {
+		return false, err
+	}
+
+	angleToExpand := s1.Angle(distance / spheroid.SphereRadius)
+	if useSphereOrSpheroid == UseSpheroid {
+		angleToExpand *= (1 + SpheroidErrorFraction)
+	}
+	if !a.BoundingCap().Expanded(angleToExpand).Intersects(b.BoundingCap()) {
+		return false, nil
+	}
 
 	aRegions, err := a.AsS2(geo.EmptyBehaviorError)
 	if err != nil {
@@ -41,10 +54,6 @@ func DWithin(
 		}
 		return false, err
 	}
-	spheroid, err := a.Spheroid()
-	if err != nil {
-		return false, err
-	}
 	maybeClosestDistance, err := distanceGeographyRegions(spheroid, useSphereOrSpheroid, aRegions, bRegions, distance)
 	if err != nil {
 		return false, err

pkg/sql/mutations/mutations.go

@@ -536,7 +536,7 @@ func postgresMutator(rng *rand.Rand, q string) string {
 var postgresStatementMutator MultiStatementMutation = func(rng *rand.Rand, stmts []tree.Statement) (mutated []tree.Statement, changed bool) {
 	for _, stmt := range stmts {
 		switch stmt := stmt.(type) {
-		case *tree.SetClusterSetting:
+		case *tree.SetClusterSetting, *tree.SetVar:
 			continue
 		case *tree.CreateTable:
 			if stmt.Interleave != nil {

pkg/sql/opt/memo/expr.go

@@ -552,6 +552,16 @@ func (s *ScanPrivate) IsCanonical() bool {
 		s.HardLimit == 0
 }
 
+// IsUnfiltered returns true if the ScanPrivate will produce all rows in the
+// table.
+func (s *ScanPrivate) IsUnfiltered(md *opt.Metadata) bool {
+	_, isPartialIndex := md.Table(s.Table).Index(s.Index).Predicate()
+	return !isPartialIndex &&
+		(s.Constraint == nil || s.Constraint.IsUnconstrained()) &&
+		s.InvertedConstraint == nil &&
+		s.HardLimit == 0
+}
+
 // IsLocking returns true if the ScanPrivate is configured to use a row-level
 // locking mode. This can be the case either because the Scan is in the scope of
 // a SELECT .. FOR [KEY] UPDATE/SHARE clause or because the Scan was configured

pkg/sql/opt/memo/multiplicity_builder.go

@@ -115,9 +115,9 @@ func deriveUnfilteredCols(in RelExpr) opt.ColSet {
 		// non-null foreign key relation - rows in kr imply rows in xy. However,
 		// the columns from xy are not output columns, so in order to see that this
 		// is the case we must bubble up non-output columns.
-		baseTable := t.Memo().Metadata().Table(t.Table)
-		_, isPartialIndex := baseTable.Index(t.Index).Predicate()
-		if t.HardLimit == 0 && t.Constraint == nil && !isPartialIndex {
+		md := t.Memo().Metadata()
+		baseTable := md.Table(t.Table)
+		if t.IsUnfiltered(md) {
 			for i, cnt := 0, baseTable.ColumnCount(); i < cnt; i++ {
 				unfilteredCols.Add(t.Table.ColumnID(i))
 			}

pkg/sql/opt/opbench/testdata/scan-lineitem.csv

@@ -1,31 +1,31 @@
 rows,num_cols,estimated,actual
-1000000,1,1170000.020000,0.834084
-2000000,1,2340000.020000,1.489845
-3000000,1,3510000.020000,2.211772
-4000000,1,4680000.020000,2.787555
-5000000,1,5850000.020000,3.727968
-6000000,1,7020000.020000,4.538592
-1000000,2,1180000.020000,0.856709
-2000000,2,2360000.020000,1.689832
-3000000,2,3540000.020000,2.624644
-4000000,2,4720000.020000,3.354893
-5000000,2,5900000.020000,4.149957
-6000000,2,7080000.020000,5.126522
-1000000,3,1190000.020000,0.948950
-2000000,3,2380000.020000,1.844214
-3000000,3,3570000.020000,2.819678
-4000000,3,4760000.020000,3.689639
-5000000,3,5950000.020000,4.598679
-6000000,3,7140000.020000,5.589131
-1000000,4,1200000.020000,0.963368
-2000000,4,2400000.020000,1.907634
-3000000,4,3600000.020000,2.819860
-4000000,4,4800000.020000,3.835917
-5000000,4,6000000.020000,4.603531
-6000000,4,7200000.020000,5.676630
-1000000,16,1320000.020000,1.951303
-2000000,16,2640000.020000,3.845009
-3000000,16,3960000.020000,5.823685
-4000000,16,5280000.020000,7.786907
-5000000,16,6600000.020000,9.746737
-6000000,16,7920000.020000,11.603873
+1000000,1,1170000.010000,0.834084
+2000000,1,2340000.010000,1.489845
+3000000,1,3510000.010000,2.211772
+4000000,1,4680000.010000,2.787555
+5000000,1,5850000.010000,3.727968
+6000000,1,7020000.010000,4.538592
+1000000,2,1180000.010000,0.856709
+2000000,2,2360000.010000,1.689832
+3000000,2,3540000.010000,2.624644
+4000000,2,4720000.010000,3.354893
+5000000,2,5900000.010000,4.149957
+6000000,2,7080000.010000,5.126522
+1000000,3,1190000.010000,0.948950
+2000000,3,2380000.010000,1.844214
+3000000,3,3570000.010000,2.819678
+4000000,3,4760000.010000,3.689639
+5000000,3,5950000.010000,4.598679
+6000000,3,7140000.010000,5.589131
+1000000,4,1200000.010000,0.963368
+2000000,4,2400000.010000,1.907634
+3000000,4,3600000.010000,2.819860
+4000000,4,4800000.010000,3.835917
+5000000,4,6000000.010000,4.603531
+6000000,4,7200000.010000,5.676630
+1000000,16,1320000.010000,1.951303
+2000000,16,2640000.010000,3.845009
+3000000,16,3960000.010000,5.823685
+4000000,16,5280000.010000,7.786907
+5000000,16,6600000.010000,9.746737
+6000000,16,7920000.010000,11.603873

pkg/sql/opt/opbench/testdata/scan-orders.csv

@@ -1,25 +1,25 @@
 rows,num_cols,estimated,actual
-250000,1,275000.020000,0.234685
-500000,1,550000.020000,0.355673
-750000,1,825000.020000,0.536125
-1000000,1,1100000.020000,0.735832
-1250000,1,1375000.020000,0.909200
-1500000,1,1650000.020000,1.036382
-250000,3,280000.020000,0.226567
-500000,3,560000.020000,0.441384
-750000,3,840000.020000,0.611747
-1000000,3,1120000.020000,0.881042
-1250000,3,1400000.020000,1.092393
-1500000,3,1680000.020000,1.314116
-250000,6,287500.020000,0.292452
-500000,6,575000.020000,0.496511
-750000,6,862500.020000,0.817809
-1000000,6,1150000.020000,1.162477
-1250000,6,1437500.020000,1.451609
-1500000,6,1725000.020000,1.760463
-250000,9,295000.020000,0.360331
-500000,9,590000.020000,0.626732
-750000,9,885000.020000,0.992292
-1000000,9,1180000.020000,1.368585
-1250000,9,1475000.020000,1.727139
-1500000,9,1770000.020000,2.056275
+250000,1,275000.010000,0.234685
+500000,1,550000.010000,0.355673
+750000,1,825000.010000,0.536125
+1000000,1,1100000.010000,0.735832
+1250000,1,1375000.010000,0.909200
+1500000,1,1650000.010000,1.036382
+250000,3,280000.010000,0.226567
+500000,3,560000.010000,0.441384
+750000,3,840000.010000,0.611747
+1000000,3,1120000.010000,0.881042
+1250000,3,1400000.010000,1.092393
+1500000,3,1680000.010000,1.314116
+250000,6,287500.010000,0.292452
+500000,6,575000.010000,0.496511
+750000,6,862500.010000,0.817809
+1000000,6,1150000.010000,1.162477
+1250000,6,1437500.010000,1.451609
+1500000,6,1725000.010000,1.760463
+250000,9,295000.010000,0.360331
+500000,9,590000.010000,0.626732
+750000,9,885000.010000,0.992292
+1000000,9,1180000.010000,1.368585
+1250000,9,1475000.010000,1.727139
+1500000,9,1770000.010000,2.056275

pkg/sql/opt/opbench/testdata/sort-lineitem.csv

@@ -1,19 +1,19 @@
 rows,num_cols,estimated,actual
-1000000,1,1588631.401386,1.467528
-2000000,1,3217262.772773,3.007519
-3000000,1,4860991.894203,5.827569
-4000000,1,6514525.515546,7.106306
-5000000,1,8175349.696421,8.982024
-6000000,1,9841983.758405,10.749726
-1000000,2,1598631.401386,1.665857
-2000000,2,3237262.772773,4.036806
-3000000,2,4890991.894203,6.109568
-4000000,2,6554525.515546,8.072787
-5000000,2,8225349.696421,9.743525
-6000000,2,9901983.758405,12.756862
-1000000,3,1608631.401386,2.786447
-2000000,3,3257262.772773,4.812669
-3000000,3,4920991.894203,7.417335
-4000000,3,6594525.515546,11.088666
-5000000,3,8275349.696421,12.703216
-6000000,3,9961983.758405,14.615791
+1000000,1,1588631.391386,1.467528
+2000000,1,3217262.762773,3.007519
+3000000,1,4860991.884203,5.827569
+4000000,1,6514525.505546,7.106306
+5000000,1,8175349.686421,8.982024
+6000000,1,9841983.748405,10.749726
+1000000,2,1598631.391386,1.665857
+2000000,2,3237262.762773,4.036806
+3000000,2,4890991.884203,6.109568
+4000000,2,6554525.505546,8.072787
+5000000,2,8225349.686421,9.743525
+6000000,2,9901983.748405,12.756862
+1000000,3,1608631.391386,2.786447
+2000000,3,3257262.762773,4.812669
+3000000,3,4920991.884203,7.417335
+4000000,3,6594525.505546,11.088666
+5000000,3,8275349.686421,12.703216
+6000000,3,9961983.748405,14.615791