Cache the computation of default columns during three-way merges.

go/libraries/doltcore/merge/merge_prolly_rows.go

@@ -130,7 +130,12 @@ func mergeProllyTableData(ctx *sql.Context, tm *TableMerger, finalSch schema.Sch
 
 	keyless := schema.IsKeyless(tm.leftSch)
 
-	pri, err := newPrimaryMerger(leftEditor, tm, valueMerger, finalSch, mergeInfo)
+	defaults, err := resolveDefaults(ctx, tm.name, finalSch, tm.leftSch)
+	if err != nil {
+		return nil, nil, err
+	}
+
+	pri, err := newPrimaryMerger(leftEditor, tm, valueMerger, finalSch, mergeInfo, defaults)
 	if err != nil {
 		return nil, nil, err
 	}
@@ -1044,15 +1049,17 @@ type primaryMerger struct {
 	tableMerger *TableMerger
 	finalSch    schema.Schema
 	mergeInfo   MergeInfo
+	defaults    []sql.Expression
 }
 
-func newPrimaryMerger(leftEditor *prolly.MutableMap, tableMerger *TableMerger, valueMerger *valueMerger, finalSch schema.Schema, mergeInfo MergeInfo) (*primaryMerger, error) {
+func newPrimaryMerger(leftEditor *prolly.MutableMap, tableMerger *TableMerger, valueMerger *valueMerger, finalSch schema.Schema, mergeInfo MergeInfo, defaults []sql.Expression) (*primaryMerger, error) {
 	return &primaryMerger{
 		mut:         leftEditor,
 		valueMerger: valueMerger,
 		tableMerger: tableMerger,
 		finalSch:    finalSch,
 		mergeInfo:   mergeInfo,
+		defaults:    defaults,
 	}, nil
 }
 
@@ -1149,13 +1156,8 @@ func (m *primaryMerger) merge(ctx *sql.Context, diff tree.ThreeWayDiff, sourceSc
 				return fmt.Errorf("cannot merge keyless tables with reordered columns")
 			}
 		} else {
-			defaults, err := resolveDefaults(ctx, m.tableMerger.name, m.finalSch, m.tableMerger.leftSch)
-			if err != nil {
-				return err
-			}
-
 			tempTupleValue, err := remapTupleWithColumnDefaults(ctx, diff.Key, newTupleValue, sourceSch.GetValueDescriptor(),
-				m.valueMerger.leftMapping, m.tableMerger, m.tableMerger.leftSch, m.finalSch, defaults, m.valueMerger.syncPool, false)
+				m.valueMerger.leftMapping, m.tableMerger, m.tableMerger.leftSch, m.finalSch, m.defaults, m.valueMerger.syncPool, false)
 			if err != nil {
 				return err
 			}

integration-tests/bats/performance-repo/.dolt/config.json

@@ -0,0 +1 @@
+{}

integration-tests/bats/performance-repo/.dolt/noms/manifest

@@ -0,0 +1 @@
+5:__DOLT__:4t8rtld9ul7l137jac93phj1o6qinq6o:8ep2ilbq0lb461mqibkqp80nd0skdend:00000000000000000000000000000000:vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv:3722

integration-tests/bats/performance-repo/.dolt/repo_state.json

@@ -0,0 +1,6 @@
+{
+  "head": "refs/heads/full",
+  "remotes": {},
+  "backups": {},
+  "branches": {}
+}

integration-tests/bats/performance.bats

@@ -0,0 +1,153 @@
+#!/usr/bin/env bats
+load $BATS_TEST_DIRNAME/helper/common.bash
+
+# This BATS test attempts to detect performance regressions when using standard workflows on large datasets.
+# Please note that this is a rough approach that is not designed to detect all performance issues, merely an extra
+# safeguard against bugs that cause large (order-of-magnitude+) regressions.
+
+# BATS_TEST_TIMEOUT is measured in seconds and is chosen to be high enough that all tests in this suite pass
+# when running on GitHub's CI, but low enough that an order-of magnitude regression will cause them to fail.
+BATS_TEST_TIMEOUT=50
+
+# This function was used to create the dolt repo used for this test. It is not run during testing.
+create_repo() {
+    dolt init
+    dolt checkout -b full
+
+    dolt sql -q 'create table t (pk int primary key, c0 text default "1", c3 text default "2", c4 text default "3", c5 text default "4", c6 text default "5", c7 text default "6");'
+    dolt commit -Am "new table t"
+
+    echo "insert into t(pk) values" > import.sql
+    for i in {1..100000}
+    do
+        echo "  ($i)," >> import.sql
+    done
+    echo "  (104857);" >> import.sql
+
+    dolt sql < import.sql
+
+    dolt add .
+    dolt commit -m "Add all rows"
+}
+
+setup() {
+    cp -r $BATS_TEST_DIRNAME/performance-repo/ $BATS_TMPDIR/dolt-repo-$$
+    cd $BATS_TMPDIR/dolt-repo-$$
+}
+
+@test "performance: merge with no schema change and no conflict" {
+    dolt checkout full
+    dolt checkout -b mod2
+    dolt reset --soft HEAD^
+
+    dolt sql -q "delete from t where pk % 2 = 0"
+
+    dolt add .
+    dolt commit -m "Add mod2 rows"
+
+    dolt checkout full
+    dolt checkout -b mod3
+    dolt reset --soft HEAD^
+
+    dolt sql -q "delete from t where pk % 3 = 0"
+
+    dolt add .
+    dolt commit -m "Add mod3 rows"
+
+    run dolt merge mod2
+    log_status_eq 0
+}
+
+@test "performance: merge with no schema change and conflict" {
+    dolt checkout full
+    dolt checkout -b mod2
+    dolt reset --soft HEAD^
+
+    dolt sql -q "delete from t where pk % 2 = 0"
+
+    dolt add .
+    dolt commit -m "Add mod2 rows"
+
+    dolt checkout full
+    dolt checkout -b mod3
+    dolt reset --soft HEAD^
+
+    dolt sql -q "delete from t where pk % 3 = 0"
+    dolt sql -q 'update t set c0 = "conflict" where pk = 91'
+
+    dolt add .
+    dolt commit -m "Add mod3 rows"
+
+    run dolt merge mod2
+    log_status_eq 1
+    [[ "$output" =~ "Merge conflict in t" ]] || false
+
+    dolt conflicts resolve --ours t
+    BATS_TEST_TIMEOUT=1
+}
+
+
+@test "performance: merge with schema change and no conflict" {
+    dolt checkout full
+    dolt checkout -b mod2
+    dolt reset --soft HEAD^
+
+    dolt sql -q "delete from t where pk % 2 = 0"
+
+    dolt add .
+    dolt commit -m "Add mod2 rows"
+
+    dolt sql -q "alter table t add column c1 int default 1"
+    dolt add .
+    dolt commit -m "Add column c1"
+
+    dolt checkout full
+    dolt checkout -b mod3
+    dolt reset --soft HEAD^
+
+    dolt sql -q "delete from t where pk % 3 = 0"
+
+    dolt add .
+    dolt commit -m "Add mod3 rows"
+
+    dolt sql -q "alter table t add column c2 int default 2"
+    dolt add .
+    dolt commit -m "Add column c2"
+
+    run dolt merge mod2
+    log_status_eq 0
+}
+
+@test "performance: merge with schema change and conflict" {
+    dolt checkout full
+    dolt checkout -b mod2
+    dolt reset --soft HEAD^
+
+    dolt sql -q "delete from t where pk % 2 = 0"
+
+    dolt add .
+    dolt commit -m "Add mod2 rows"
+
+    dolt sql -q "alter table t add column c1 int default 1"
+    dolt add .
+    dolt commit -m "Add column c1"
+
+    dolt checkout full
+    dolt checkout -b mod3
+    dolt reset --soft HEAD^
+
+    dolt sql -q "delete from t where pk % 3 = 0"
+    dolt sql -q 'update t set c0 = "conflict" where pk = 91'
+
+    dolt add .
+    dolt commit -m "Add mod3 rows"
+
+    dolt sql -q "alter table t add column c2 int default 2"
+    dolt add .
+    dolt commit -m "Add column c2"
+
+    run dolt merge mod2
+    log_status_eq 1
+
+    dolt conflicts resolve --ours t
+}