Merge #39375 #39420 #39430 #39461

39375: exec: add null handling to default sorter r=rohany a=rohany

Addresses part of #36880.

Release note: None

39420: roachtest: fix pgjdbc tests r=rafiss a=rafiss

The pgjdbc tests have two types of failures: one referred to as
"failures" and the other one as "errors." We were only counting the
former before. Now there are many more test failures that should be
tracked in the blacklist.

touches #39406 

Release note: None

39430: workload/ycsb: implement read-modify-write r=jeffrey-xiao a=jeffrey-xiao

Release note: None

39461: internal/sqlsmith: add vectorize option r=mjibson a=mjibson

This option attempts to limits query generation such that vectorized
execution will occur. Still a lot of work to do but this is a good start.

Release note: None

Co-authored-by: Rohan Yadav <[email protected]>
Co-authored-by: Rafi Shamim <[email protected]>
Co-authored-by: Jeffrey Xiao <[email protected]>
Co-authored-by: Matt Jibson <[email protected]>

pkg/cmd/roachtest/hibernate.go

@@ -179,6 +179,8 @@ func registerHibernate(r *testRegistry) {
 		var passUnexpectedCount, passExpectedCount, notRunCount int
 		// Put all the results in a giant map of [testname]result
 		results := make(map[string]string)
+		// Put all issue hints in a map of [testname]issue
+		allIssueHints := make(map[string]string)
 		// Current failures are any tests that reported as failed, regardless of if
 		// they were expected or not.
 		var currentFailures, allTests []string
@@ -207,10 +209,13 @@ func registerHibernate(r *testRegistry) {
 			if err != nil {
 				t.Fatal(err)
 			}
-			tests, passed, err := extractFailureFromJUnitXML(fileOutput)
+			tests, passed, issueHints, err := extractFailureFromJUnitXML(fileOutput)
 			if err != nil {
 				t.Fatal(err)
 			}
+			for testName, issue := range issueHints {
+				allIssueHints[testName] = issue
+			}
 			for i, test := range tests {
 				// There is at least a single test that's run twice, so if we already
 				// have a result, skip it.
@@ -219,6 +224,9 @@ func registerHibernate(r *testRegistry) {
 				}
 				allTests = append(allTests, test)
 				issue, expectedFailure := expectedFailures[test]
+				if len(issue) == 0 || issue == "unknown" {
+					issue = issueHints[test]
+				}
 				pass := passed[i]
 				switch {
 				case pass && !expectedFailure:
@@ -236,7 +244,8 @@ func registerHibernate(r *testRegistry) {
 					failExpectedCount++
 					currentFailures = append(currentFailures, test)
 				case !pass && !expectedFailure:
-					results[test] = fmt.Sprintf("--- FAIL: %s (unexpected)", test)
+					results[test] = fmt.Sprintf("--- FAIL: %s - %s (unexpected)",
+						test, maybeAddGithubLink(issue))
 					failUnexpectedCount++
 					currentFailures = append(currentFailures, test)
 				}
@@ -298,6 +307,9 @@ func registerHibernate(r *testRegistry) {
 			fmt.Fprintf(&b, "var %s = blacklist{\n", blacklistName)
 			for _, test := range currentFailures {
 				issue := expectedFailures[test]
+				if len(issue) == 0 || issue == "unknown" {
+					issue = allIssueHints[test]
+				}
 				if len(issue) == 0 {
 					issue = "unknown"
 				}

pkg/cmd/roachtest/java_helpers.go

@@ -13,16 +13,30 @@ package main
 import (
 	"encoding/xml"
 	"fmt"
+	"regexp"
 )
 
-func extractFailureFromJUnitXML(contents []byte) ([]string, []bool, error) {
+var issueRegexp = regexp.MustCompile(`See: https://github.com/cockroachdb/cockroach/issues/(\d+)`)
+
+// extractFailureFromJUnitXML parses an XML report to find all failed tests. The
+// return values are:
+// - slice of all test names.
+// - slice of bool for each test, with true indicating pass.
+// - map from name of a failed test to a github issue that explains the failure,
+//   if the error message contained a reference to an issue.
+// - error if there was a problem parsing the XML.
+func extractFailureFromJUnitXML(contents []byte) ([]string, []bool, map[string]string, error) {
 	type Failure struct {
 		Message string `xml:"message,attr"`
 	}
+	type Error struct {
+		Message string `xml:"message,attr"`
+	}
 	type TestCase struct {
 		Name      string  `xml:"name,attr"`
 		ClassName string  `xml:"classname,attr"`
 		Failure   Failure `xml:"failure,omitempty"`
+		Error     Error   `xml:"error,omitempty"`
 	}
 	type TestSuite struct {
 		XMLName   xml.Name   `xml:"testsuite"`
@@ -33,15 +47,31 @@ func extractFailureFromJUnitXML(contents []byte) ([]string, []bool, error) {
 	_ = testSuite.XMLName
 
 	if err := xml.Unmarshal(contents, &testSuite); err != nil {
-		return nil, nil, err
+		return nil, nil, nil, err
 	}
 
 	var tests []string
 	var passed []bool
+	var failedTestToIssue = make(map[string]string)
 	for _, testCase := range testSuite.TestCases {
-		tests = append(tests, fmt.Sprintf("%s.%s", testCase.ClassName, testCase.Name))
-		passed = append(passed, len(testCase.Failure.Message) == 0)
+		testName := fmt.Sprintf("%s.%s", testCase.ClassName, testCase.Name)
+		testPassed := len(testCase.Failure.Message) == 0 && len(testCase.Error.Message) == 0
+		tests = append(tests, testName)
+		passed = append(passed, testPassed)
+		if !testPassed {
+			message := testCase.Failure.Message
+			if len(message) == 0 {
+				message = testCase.Error.Message
+			}
+
+			issue := "unknown"
+			match := issueRegexp.FindStringSubmatch(message)
+			if match != nil {
+				issue = match[1]
+			}
+			failedTestToIssue[testName] = issue
+		}
 	}
 
-	return tests, passed, nil
+	return tests, passed, failedTestToIssue, nil
 }

pkg/cmd/roachtest/pgjdbc.go

@@ -165,6 +165,8 @@ func registerPgjdbc(r *testRegistry) {
 		var passUnexpectedCount, passExpectedCount, notRunCount int
 		// Put all the results in a giant map of [testname]result
 		results := make(map[string]string)
+		// Put all issue hints in a map of [testname]issue
+		allIssueHints := make(map[string]string)
 		// Current failures are any tests that reported as failed, regardless of if
 		// they were expected or not.
 		var currentFailures, allTests []string
@@ -193,10 +195,13 @@ func registerPgjdbc(r *testRegistry) {
 			if err != nil {
 				t.Fatal(err)
 			}
-			tests, passed, err := extractFailureFromJUnitXML(fileOutput)
+			tests, passed, issueHints, err := extractFailureFromJUnitXML(fileOutput)
 			if err != nil {
 				t.Fatal(err)
 			}
+			for testName, issue := range issueHints {
+				allIssueHints[testName] = issue
+			}
 			for i, test := range tests {
 				// There is at least a single test that's run twice, so if we already
 				// have a result, skip it.
@@ -205,6 +210,9 @@ func registerPgjdbc(r *testRegistry) {
 				}
 				allTests = append(allTests, test)
 				issue, expectedFailure := expectedFailures[test]
+				if len(issue) == 0 || issue == "unknown" {
+					issue = issueHints[test]
+				}
 				pass := passed[i]
 				switch {
 				case pass && !expectedFailure:
@@ -222,7 +230,8 @@ func registerPgjdbc(r *testRegistry) {
 					failExpectedCount++
 					currentFailures = append(currentFailures, test)
 				case !pass && !expectedFailure:
-					results[test] = fmt.Sprintf("--- FAIL: %s (unexpected)", test)
+					results[test] = fmt.Sprintf("--- FAIL: %s - %s (unexpected)",
+						test, maybeAddGithubLink(issue))
 					failUnexpectedCount++
 					currentFailures = append(currentFailures, test)
 				}
@@ -283,6 +292,9 @@ func registerPgjdbc(r *testRegistry) {
 			fmt.Fprintf(&b, "var %s = blacklist{\n", blacklistName)
 			for _, test := range currentFailures {
 				issue := expectedFailures[test]
+				if len(issue) == 0 || issue == "unknown" {
+					issue = allIssueHints[test]
+				}
 				if len(issue) == 0 {
 					issue = "unknown"
 				}