[GitHub] [spark] Ngone51 commented on a change in pull request #29270: [SPARK-32466][TEST][SQL] Add PlanStabilitySuite to detect SparkPlan regression
Ngone51 commented on a change in pull request #29270:
URL: https://github.com/apache/spark/pull/29270#discussion_r476196472
##
File path: sql/core/src/test/scala/org/apache/spark/sql/PlanStabilitySuite.scala
##
@@ -0,0 +1,327 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ *http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.sql
+
+import java.io.File
+import java.nio.charset.StandardCharsets
+
+import scala.collection.mutable
+
+import org.apache.commons.io.FileUtils
+
+import org.apache.spark.sql.catalyst.expressions.AttributeSet
+import org.apache.spark.sql.catalyst.util._
+import org.apache.spark.sql.execution._
+import org.apache.spark.sql.execution.adaptive.DisableAdaptiveExecutionSuite
+import org.apache.spark.sql.execution.exchange.{Exchange, ReusedExchangeExec}
+import org.apache.spark.sql.internal.SQLConf
+import org.apache.spark.tags.ExtendedSQLTest
+
+// scalastyle:off line.size.limit
+/**
+ * Check that TPC-DS SparkPlans don't change.
+ * If there are plan differences, the error message looks like this:
+ * Plans did not match:
+ * last approved simplified plan:
/path/to/tpcds-plan-stability/approved-plans-xxx/q1/simplified.txt
+ * last approved explain plan:
/path/to/tpcds-plan-stability/approved-plans-xxx/q1/explain.txt
+ * [last approved simplified plan]
+ *
+ * actual simplified plan: /path/to/tmp/q1.actual.simplified.txt
+ * actual explain plan: /path/to/tmp/q1.actual.explain.txt
+ * [actual simplified plan]
+ *
+ * The explain files are saved to help debug later, they are not checked. Only
the simplified
+ * plans are checked (by string comparison).
+ *
+ *
+ * To run the entire test suite:
+ * {{{
+ * build/sbt "sql/test-only *PlanStability[WithStats]Suite"
+ * }}}
+ *
+ * To run a single test file upon change:
+ * {{{
+ * build/sbt "sql/test-only *PlanStability[WithStats]Suite -- -z
(tpcds-v1.4/q49)"
+ * }}}
+ *
+ * To re-generate golden files for entire suite, run:
+ * {{{
+ * SPARK_GENERATE_GOLDEN_FILES=1 build/sbt "sql/test-only
*PlanStability[WithStats]Suite"
+ * }}}
+ *
+ * To re-generate golden file for a single test, run:
+ * {{{
+ * SPARK_GENERATE_GOLDEN_FILES=1 build/sbt "sql/test-only
*PlanStability[WithStats]Suite -- -z (tpcds-v1.4/q49)"
+ * }}}
+ */
+// scalastyle:on line.size.limit
+trait PlanStabilitySuite extends TPCDSBase with DisableAdaptiveExecutionSuite {
+
+ private val originalMaxToStringFields = conf.maxToStringFields
+
+ override def beforeAll(): Unit = {
+conf.setConf(SQLConf.MAX_TO_STRING_FIELDS, Int.MaxValue)
+super.beforeAll()
+ }
+
+ override def afterAll(): Unit = {
+super.afterAll()
+conf.setConf(SQLConf.MAX_TO_STRING_FIELDS, originalMaxToStringFields)
+ }
+
+ private val regenerateGoldenFiles: Boolean =
System.getenv("SPARK_GENERATE_GOLDEN_FILES") == "1"
+
+ protected val baseResourcePath = {
+// use the same way as `SQLQueryTestSuite` to get the resource path
+java.nio.file.Paths.get("src", "test", "resources",
"tpcds-plan-stability").toFile
+ }
+
+ private val referenceRegex = "#\\d+".r
+ private val normalizeRegex = "#\\d+L?".r
+
+ def goldenFilePath: String
+
+ private def getDirForTest(name: String): File = {
+new File(goldenFilePath, name)
+ }
+
+ private def isApproved(dir: File, actualSimplifiedPlan: String): Boolean = {
+val file = new File(dir, "simplified.txt")
+val approved = FileUtils.readFileToString(file, StandardCharsets.UTF_8)
+approved == actualSimplifiedPlan
+ }
+
+ /**
+ * Serialize and save this SparkPlan.
+ * The resulting file is used by [[checkWithApproved]] to check stability.
+ *
+ * @param planthe SparkPlan
+ * @param namethe name of the query
+ * @param explain the full explain output; this is saved to help debug later
as the simplified
+ *plan is not too useful for debugging
+ */
+ private def generateApprovedPlanFile(plan: SparkPlan, name: String, explain:
String): Unit = {
+val dir = getDirForTest(name)
+val simplified = getSimplifiedPlan(plan)
+val foundMatch = dir.exists() && isApproved(dir, simplified)
+
+if (!foundMatch) {
+ FileUtils.deleteDirectory(dir)
+ assert(dir.mkdirs())
+
+ val file = new File(dir,
[GitHub] [spark] Ngone51 commented on a change in pull request #29270: [SPARK-32466][TEST][SQL] Add PlanStabilitySuite to detect SparkPlan regression
Ngone51 commented on a change in pull request #29270:
URL: https://github.com/apache/spark/pull/29270#discussion_r476193874
##
File path: sql/core/src/test/scala/org/apache/spark/sql/PlanStabilitySuite.scala
##
@@ -0,0 +1,327 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ *http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.sql
+
+import java.io.File
+import java.nio.charset.StandardCharsets
+
+import scala.collection.mutable
+
+import org.apache.commons.io.FileUtils
+
+import org.apache.spark.sql.catalyst.expressions.AttributeSet
+import org.apache.spark.sql.catalyst.util._
+import org.apache.spark.sql.execution._
+import org.apache.spark.sql.execution.adaptive.DisableAdaptiveExecutionSuite
+import org.apache.spark.sql.execution.exchange.{Exchange, ReusedExchangeExec}
+import org.apache.spark.sql.internal.SQLConf
+import org.apache.spark.tags.ExtendedSQLTest
+
+// scalastyle:off line.size.limit
+/**
+ * Check that TPC-DS SparkPlans don't change.
+ * If there are plan differences, the error message looks like this:
+ * Plans did not match:
+ * last approved simplified plan:
/path/to/tpcds-plan-stability/approved-plans-xxx/q1/simplified.txt
+ * last approved explain plan:
/path/to/tpcds-plan-stability/approved-plans-xxx/q1/explain.txt
+ * [last approved simplified plan]
+ *
+ * actual simplified plan: /path/to/tmp/q1.actual.simplified.txt
+ * actual explain plan: /path/to/tmp/q1.actual.explain.txt
+ * [actual simplified plan]
+ *
+ * The explain files are saved to help debug later, they are not checked. Only
the simplified
+ * plans are checked (by string comparison).
+ *
+ *
+ * To run the entire test suite:
+ * {{{
+ * build/sbt "sql/test-only *PlanStability[WithStats]Suite"
+ * }}}
+ *
+ * To run a single test file upon change:
+ * {{{
+ * build/sbt "sql/test-only *PlanStability[WithStats]Suite -- -z
(tpcds-v1.4/q49)"
+ * }}}
+ *
+ * To re-generate golden files for entire suite, run:
+ * {{{
+ * SPARK_GENERATE_GOLDEN_FILES=1 build/sbt "sql/test-only
*PlanStability[WithStats]Suite"
+ * }}}
+ *
+ * To re-generate golden file for a single test, run:
+ * {{{
+ * SPARK_GENERATE_GOLDEN_FILES=1 build/sbt "sql/test-only
*PlanStability[WithStats]Suite -- -z (tpcds-v1.4/q49)"
+ * }}}
+ */
+// scalastyle:on line.size.limit
+trait PlanStabilitySuite extends TPCDSBase with DisableAdaptiveExecutionSuite {
+
+ private val originalMaxToStringFields = conf.maxToStringFields
+
+ override def beforeAll(): Unit = {
+conf.setConf(SQLConf.MAX_TO_STRING_FIELDS, Int.MaxValue)
+super.beforeAll()
+ }
+
+ override def afterAll(): Unit = {
+super.afterAll()
+conf.setConf(SQLConf.MAX_TO_STRING_FIELDS, originalMaxToStringFields)
+ }
+
+ private val regenerateGoldenFiles: Boolean =
System.getenv("SPARK_GENERATE_GOLDEN_FILES") == "1"
+
+ protected val baseResourcePath = {
+// use the same way as `SQLQueryTestSuite` to get the resource path
+java.nio.file.Paths.get("src", "test", "resources",
"tpcds-plan-stability").toFile
+ }
+
+ private val referenceRegex = "#\\d+".r
+ private val normalizeRegex = "#\\d+L?".r
+
+ def goldenFilePath: String
+
+ private def getDirForTest(name: String): File = {
+new File(goldenFilePath, name)
+ }
+
+ private def isApproved(dir: File, actualSimplifiedPlan: String): Boolean = {
+val file = new File(dir, "simplified.txt")
+val approved = FileUtils.readFileToString(file, StandardCharsets.UTF_8)
+approved == actualSimplifiedPlan
+ }
+
+ /**
+ * Serialize and save this SparkPlan.
+ * The resulting file is used by [[checkWithApproved]] to check stability.
+ *
+ * @param planthe SparkPlan
+ * @param namethe name of the query
+ * @param explain the full explain output; this is saved to help debug later
as the simplified
+ *plan is not too useful for debugging
+ */
+ private def generateApprovedPlanFile(plan: SparkPlan, name: String, explain:
String): Unit = {
Review comment:
I can rename it to `generateGoldenFile` since we prefer `expected`
compares to `approved`.
This is an automated message from the Apache Git Service.
To respond to the message,
[GitHub] [spark] Ngone51 commented on a change in pull request #29270: [SPARK-32466][TEST][SQL] Add PlanStabilitySuite to detect SparkPlan regression
Ngone51 commented on a change in pull request #29270:
URL: https://github.com/apache/spark/pull/29270#discussion_r476191815
##
File path: sql/core/src/test/scala/org/apache/spark/sql/PlanStabilitySuite.scala
##
@@ -0,0 +1,327 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ *http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.sql
+
+import java.io.File
+import java.nio.charset.StandardCharsets
+
+import scala.collection.mutable
+
+import org.apache.commons.io.FileUtils
+
+import org.apache.spark.sql.catalyst.expressions.AttributeSet
+import org.apache.spark.sql.catalyst.util._
+import org.apache.spark.sql.execution._
+import org.apache.spark.sql.execution.adaptive.DisableAdaptiveExecutionSuite
+import org.apache.spark.sql.execution.exchange.{Exchange, ReusedExchangeExec}
+import org.apache.spark.sql.internal.SQLConf
+import org.apache.spark.tags.ExtendedSQLTest
+
+// scalastyle:off line.size.limit
+/**
+ * Check that TPC-DS SparkPlans don't change.
+ * If there are plan differences, the error message looks like this:
+ * Plans did not match:
+ * last approved simplified plan:
/path/to/tpcds-plan-stability/approved-plans-xxx/q1/simplified.txt
+ * last approved explain plan:
/path/to/tpcds-plan-stability/approved-plans-xxx/q1/explain.txt
+ * [last approved simplified plan]
+ *
+ * actual simplified plan: /path/to/tmp/q1.actual.simplified.txt
+ * actual explain plan: /path/to/tmp/q1.actual.explain.txt
+ * [actual simplified plan]
+ *
+ * The explain files are saved to help debug later, they are not checked. Only
the simplified
+ * plans are checked (by string comparison).
+ *
+ *
+ * To run the entire test suite:
+ * {{{
+ * build/sbt "sql/test-only *PlanStability[WithStats]Suite"
+ * }}}
+ *
+ * To run a single test file upon change:
+ * {{{
+ * build/sbt "sql/test-only *PlanStability[WithStats]Suite -- -z
(tpcds-v1.4/q49)"
+ * }}}
+ *
+ * To re-generate golden files for entire suite, run:
+ * {{{
+ * SPARK_GENERATE_GOLDEN_FILES=1 build/sbt "sql/test-only
*PlanStability[WithStats]Suite"
+ * }}}
+ *
+ * To re-generate golden file for a single test, run:
+ * {{{
+ * SPARK_GENERATE_GOLDEN_FILES=1 build/sbt "sql/test-only
*PlanStability[WithStats]Suite -- -z (tpcds-v1.4/q49)"
+ * }}}
+ */
+// scalastyle:on line.size.limit
+trait PlanStabilitySuite extends TPCDSBase with DisableAdaptiveExecutionSuite {
+
+ private val originalMaxToStringFields = conf.maxToStringFields
+
+ override def beforeAll(): Unit = {
+conf.setConf(SQLConf.MAX_TO_STRING_FIELDS, Int.MaxValue)
+super.beforeAll()
+ }
+
+ override def afterAll(): Unit = {
+super.afterAll()
+conf.setConf(SQLConf.MAX_TO_STRING_FIELDS, originalMaxToStringFields)
+ }
+
+ private val regenerateGoldenFiles: Boolean =
System.getenv("SPARK_GENERATE_GOLDEN_FILES") == "1"
+
+ protected val baseResourcePath = {
+// use the same way as `SQLQueryTestSuite` to get the resource path
+java.nio.file.Paths.get("src", "test", "resources",
"tpcds-plan-stability").toFile
+ }
+
+ private val referenceRegex = "#\\d+".r
+ private val normalizeRegex = "#\\d+L?".r
+
+ def goldenFilePath: String
+
+ private def getDirForTest(name: String): File = {
+new File(goldenFilePath, name)
+ }
+
+ private def isApproved(dir: File, actualSimplifiedPlan: String): Boolean = {
+val file = new File(dir, "simplified.txt")
+val approved = FileUtils.readFileToString(file, StandardCharsets.UTF_8)
+approved == actualSimplifiedPlan
Review comment:
Sounds reasonable.
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] Ngone51 commented on a change in pull request #29270: [SPARK-32466][TEST][SQL] Add PlanStabilitySuite to detect SparkPlan regression
Ngone51 commented on a change in pull request #29270: URL: https://github.com/apache/spark/pull/29270#discussion_r475307366 ## File path: sql/core/src/test/resources/tpcds-plan-stability/approved-plans-modified/q10.sf100/explain.txt ## @@ -0,0 +1,286 @@ +== Physical Plan == +TakeOrderedAndProject (52) ++- * HashAggregate (51) + +- Exchange (50) + +- * HashAggregate (49) + +- * Project (48) ++- * BroadcastHashJoin Inner BuildLeft (47) + :- BroadcastExchange (43) + : +- * Project (42) + : +- * BroadcastHashJoin Inner BuildRight (41) + ::- * Project (35) + :: +- SortMergeJoin LeftSemi (34) + :: :- SortMergeJoin LeftSemi (25) + :: : :- * Sort (5) + :: : : +- Exchange (4) + :: : : +- * Filter (3) + :: : :+- * ColumnarToRow (2) + :: : : +- Scan parquet default.customer (1) + :: : +- * Sort (24) + :: : +- Exchange (23) + :: :+- Union (22) + :: : :- * Project (15) + :: : : +- * BroadcastHashJoin Inner BuildRight (14) + :: : : :- * Filter (8) + :: : : : +- * ColumnarToRow (7) + :: : : : +- Scan parquet default.web_sales (6) + :: : : +- BroadcastExchange (13) + :: : :+- * Project (12) + :: : : +- * Filter (11) + :: : : +- * ColumnarToRow (10) + :: : : +- Scan parquet default.date_dim (9) + :: : +- * Project (21) + :: : +- * BroadcastHashJoin Inner BuildRight (20) + :: : :- * Filter (18) + :: : : +- * ColumnarToRow (17) + :: : : +- Scan parquet default.catalog_sales (16) + :: : +- ReusedExchange (19) + :: +- * Sort (33) + ::+- Exchange (32) + :: +- * Project (31) + :: +- * BroadcastHashJoin Inner BuildRight (30) + :: :- * Filter (28) + :: : +- * ColumnarToRow (27) + :: : +- Scan parquet default.store_sales (26) + :: +- ReusedExchange (29) + :+- BroadcastExchange (40) + : +- * Project (39) + : +- * Filter (38) + : +- * ColumnarToRow (37) + :+- Scan parquet default.customer_address (36) + +- * Filter (46) + +- * ColumnarToRow (45) + +- Scan parquet default.customer_demographics (44) + + +(1) Scan parquet default.customer +Output [3]: [c_customer_sk#1, c_current_cdemo_sk#2, c_current_addr_sk#3] +Batched: true +Location: InMemoryFileIndex [file:/Users/yi.wu/IdeaProjects/spark/sql/core/spark-warehouse/org.apache.spark.sql.TPCDSModifiedPlanStabilityWithStatsSuite/customer] Review comment: Oh, I see. Thanks for pointing it out. I'll make a follow-up soon. This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org - To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] Ngone51 commented on a change in pull request #29270: [SPARK-32466][TEST][SQL] Add PlanStabilitySuite to detect SparkPlan regression
Ngone51 commented on a change in pull request #29270:
URL: https://github.com/apache/spark/pull/29270#discussion_r471236995
##
File path: sql/core/src/test/scala/org/apache/spark/sql/TPCDSBase.scala
##
@@ -256,4 +285,39 @@ trait TPCDSSchema {
|${options.mkString("\n")}
""".stripMargin)
}
+
+ private val originalCBCEnabled = conf.cboEnabled
+ private val originalPlanStatsEnabled = conf.planStatsEnabled
+ private val originalJoinReorderEnabled = conf.joinReorderEnabled
+
+ override def beforeAll(): Unit = {
+super.beforeAll()
+if (injectStats) {
+ // Sets configurations for enabling the optimization rules that
+ // exploit data statistics.
+ conf.setConf(SQLConf.CBO_ENABLED, true)
+ conf.setConf(SQLConf.PLAN_STATS_ENABLED, true)
+ conf.setConf(SQLConf.JOIN_REORDER_ENABLED, true)
Review comment:
The configurations have been extracted to the base class(`TPCDSBase`).
If we change it here, I think it will also take effect for the
`TPCDSQuerySuite`.
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] Ngone51 commented on a change in pull request #29270: [SPARK-32466][TEST][SQL] Add PlanStabilitySuite to detect SparkPlan regression
Ngone51 commented on a change in pull request #29270:
URL: https://github.com/apache/spark/pull/29270#discussion_r471203868
##
File path: sql/core/src/test/scala/org/apache/spark/sql/TPCDSBase.scala
##
@@ -256,4 +285,39 @@ trait TPCDSSchema {
|${options.mkString("\n")}
""".stripMargin)
}
+
+ private val originalCBCEnabled = conf.cboEnabled
+ private val originalPlanStatsEnabled = conf.planStatsEnabled
+ private val originalJoinReorderEnabled = conf.joinReorderEnabled
+
+ override def beforeAll(): Unit = {
+super.beforeAll()
+if (injectStats) {
+ // Sets configurations for enabling the optimization rules that
+ // exploit data statistics.
+ conf.setConf(SQLConf.CBO_ENABLED, true)
+ conf.setConf(SQLConf.PLAN_STATS_ENABLED, true)
+ conf.setConf(SQLConf.JOIN_REORDER_ENABLED, true)
Review comment:
hmm..I'm not sure about this. but it seems `preferSortMergeJoin` is
enabled by default? Or you mean we should test both `preferSortMergeJoin`
enabled and disabled?
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] Ngone51 commented on a change in pull request #29270: [SPARK-32466][TEST][SQL] Add PlanStabilitySuite to detect SparkPlan regression
Ngone51 commented on a change in pull request #29270:
URL: https://github.com/apache/spark/pull/29270#discussion_r470368611
##
File path: sql/core/src/test/scala/org/apache/spark/sql/TPCDSBase.scala
##
@@ -298,12 +302,22 @@ trait TPCDSBase extends SharedSparkSession {
tableNames.foreach { tableName =>
createTable(spark, tableName)
if (injectStats) {
-// To simulate plan generation on actual TPCDS data, injects data
stats here
+// To simulate plan generation on actual TPC-DS data, injects data
stats here
spark.sessionState.catalog.alterTableStats(
TableIdentifier(tableName),
Some(TPCDSTableStats.sf100TableStats(tableName)))
}
}
}
+ override def afterAll(): Unit = {
+conf.setConf(SQLConf.CBO_ENABLED, originalCBCEnabled)
+conf.setConf(SQLConf.PLAN_STATS_ENABLED, originalPlanStatsEnabled)
+conf.setConf(SQLConf.JOIN_REORDER_ENABLED, originalJoinReorderEnabled)
+tableNames.foreach { tableName =>
+ spark.sessionState.catalog.alterTableStats(TableIdentifier(tableName),
None)
Review comment:
Yes sure.
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] Ngone51 commented on a change in pull request #29270: [SPARK-32466][TEST][SQL] Add PlanStabilitySuite to detect SparkPlan regression
Ngone51 commented on a change in pull request #29270:
URL: https://github.com/apache/spark/pull/29270#discussion_r470368540
##
File path: sql/core/src/test/scala/org/apache/spark/sql/PlanStabilitySuite.scala
##
@@ -0,0 +1,335 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ *http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.sql
+
+import java.io.File
+import java.nio.charset.StandardCharsets
+
+import scala.collection.mutable
+
+import org.apache.commons.io.FileUtils
+
+import org.apache.spark.sql.catalyst.expressions.AttributeSet
+import org.apache.spark.sql.catalyst.util._
+import org.apache.spark.sql.execution._
+import org.apache.spark.sql.execution.adaptive.DisableAdaptiveExecutionSuite
+import org.apache.spark.sql.execution.exchange.{Exchange, ReusedExchangeExec}
+import org.apache.spark.sql.internal.SQLConf
+import org.apache.spark.tags.ExtendedSQLTest
+
+// scalastyle:off line.size.limit
+/**
+ * Check that TPC-DS SparkPlans don't change.
+ * If there are plan differences, the error message looks like this:
+ * Plans did not match:
+ * last approved simplified plan:
/path/to/tpcds-plan-stability/approved-plans-xxx/q1/simplified.txt
+ * last approved explain plan:
/path/to/tpcds-plan-stability/approved-plans-xxx/q1/explain.txt
+ * [last approved simplified plan]
+ *
+ * actual simplified plan: /path/to/tmp/q1.actual.simplified.txt
+ * actual explain plan: /path/to/tmp/q1.actual.explain.txt
+ * [actual simplified plan]
+ *
+ * The explain files are saved to help debug later, they are not checked. Only
the simplified
+ * plans are checked (by string comparison).
+ *
+ *
+ * To run the entire test suite:
+ * {{{
+ * build/sbt "sql/test-only *PlanStability[WithStats]Suite"
+ * }}}
+ *
+ * To run a single test file upon change:
+ * {{{
+ * build/sbt "sql/test-only *PlanStability[WithStats]Suite -- -z
(tpcds-v1.4/q49)"
+ * }}}
+ *
+ * To re-generate golden files for entire suite, run:
+ * {{{
+ * SPARK_GENERATE_GOLDEN_FILES=1 build/sbt "sql/test-only
*PlanStability[WithStats]Suite"
+ * }}}
+ *
+ * To re-generate golden file for a single test, run:
+ * {{{
+ * SPARK_GENERATE_GOLDEN_FILES=1 build/sbt "sql/test-only
*PlanStability[WithStats]Suite -- -z (tpcds-v1.4/q49)"
+ * }}}
+ */
+// scalastyle:on line.size.limit
+trait PlanStabilitySuite extends TPCDSBase with DisableAdaptiveExecutionSuite {
+
+ private val originalMaxToStringFields = conf.maxToStringFields
+
+ override def beforeAll(): Unit = {
+conf.setConf(SQLConf.MAX_TO_STRING_FIELDS, Int.MaxValue)
+super.beforeAll()
+ }
+
+ override def afterAll(): Unit = {
+super.afterAll()
+conf.setConf(SQLConf.MAX_TO_STRING_FIELDS, originalMaxToStringFields)
+ }
+
+ private val regenerateGoldenFiles: Boolean =
System.getenv("SPARK_GENERATE_GOLDEN_FILES") == "1"
+
+ protected val baseResourcePath = {
+// use the same way as `SQLQueryTestSuite` to get the resource path
+java.nio.file.Paths.get("src", "test", "resources",
"tpcds-plan-stability").toFile
+ }
+
+ def goldenFilePath: String
+
+ private def getDirForTest(name: String): File = {
+new File(goldenFilePath, name)
+ }
+
+ private def isApproved(dir: File, actualSimplifiedPlan: String): Boolean = {
+val file = new File(dir, "simplified.txt")
+val approved = FileUtils.readFileToString(file, StandardCharsets.UTF_8)
+approved == actualSimplifiedPlan
+ }
+
+ /**
+ * Serialize and save this SparkPlan.
+ * The resulting file is used by [[checkWithApproved]] to check stability.
+ *
+ * @param planthe SparkPlan
+ * @param namethe name of the query
+ * @param explain the full explain output; this is saved to help debug later
as the simplified
+ *plan is not too useful for debugging
+ */
+ private def generateApprovedPlanFile(plan: SparkPlan, name: String, explain:
String): Unit = {
+val dir = getDirForTest(name)
+val simplified = getSimplifiedPlan(plan)
+val foundMatch = dir.exists() && isApproved(dir, simplified)
+
+if (!foundMatch) {
+ FileUtils.deleteDirectory(dir)
+ assert(dir.mkdirs())
+
+ val file = new File(dir, "simplified.txt")
+ FileUtils.writeStringToFile(file, simplified,
[GitHub] [spark] Ngone51 commented on a change in pull request #29270: [SPARK-32466][TEST][SQL] Add PlanStabilitySuite to detect SparkPlan regression
Ngone51 commented on a change in pull request #29270:
URL: https://github.com/apache/spark/pull/29270#discussion_r470368356
##
File path: sql/core/src/test/scala/org/apache/spark/sql/PlanStabilitySuite.scala
##
@@ -0,0 +1,335 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ *http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.sql
+
+import java.io.File
+import java.nio.charset.StandardCharsets
+
+import scala.collection.mutable
+
+import org.apache.commons.io.FileUtils
+
+import org.apache.spark.sql.catalyst.expressions.AttributeSet
+import org.apache.spark.sql.catalyst.util._
+import org.apache.spark.sql.execution._
+import org.apache.spark.sql.execution.adaptive.DisableAdaptiveExecutionSuite
+import org.apache.spark.sql.execution.exchange.{Exchange, ReusedExchangeExec}
+import org.apache.spark.sql.internal.SQLConf
+import org.apache.spark.tags.ExtendedSQLTest
+
+// scalastyle:off line.size.limit
+/**
+ * Check that TPC-DS SparkPlans don't change.
+ * If there are plan differences, the error message looks like this:
+ * Plans did not match:
+ * last approved simplified plan:
/path/to/tpcds-plan-stability/approved-plans-xxx/q1/simplified.txt
+ * last approved explain plan:
/path/to/tpcds-plan-stability/approved-plans-xxx/q1/explain.txt
+ * [last approved simplified plan]
+ *
+ * actual simplified plan: /path/to/tmp/q1.actual.simplified.txt
+ * actual explain plan: /path/to/tmp/q1.actual.explain.txt
+ * [actual simplified plan]
+ *
+ * The explain files are saved to help debug later, they are not checked. Only
the simplified
+ * plans are checked (by string comparison).
+ *
+ *
+ * To run the entire test suite:
+ * {{{
+ * build/sbt "sql/test-only *PlanStability[WithStats]Suite"
+ * }}}
+ *
+ * To run a single test file upon change:
+ * {{{
+ * build/sbt "sql/test-only *PlanStability[WithStats]Suite -- -z
(tpcds-v1.4/q49)"
+ * }}}
+ *
+ * To re-generate golden files for entire suite, run:
+ * {{{
+ * SPARK_GENERATE_GOLDEN_FILES=1 build/sbt "sql/test-only
*PlanStability[WithStats]Suite"
+ * }}}
+ *
+ * To re-generate golden file for a single test, run:
+ * {{{
+ * SPARK_GENERATE_GOLDEN_FILES=1 build/sbt "sql/test-only
*PlanStability[WithStats]Suite -- -z (tpcds-v1.4/q49)"
+ * }}}
+ */
+// scalastyle:on line.size.limit
+trait PlanStabilitySuite extends TPCDSBase with DisableAdaptiveExecutionSuite {
+
+ private val originalMaxToStringFields = conf.maxToStringFields
+
+ override def beforeAll(): Unit = {
+conf.setConf(SQLConf.MAX_TO_STRING_FIELDS, Int.MaxValue)
+super.beforeAll()
+ }
+
+ override def afterAll(): Unit = {
+super.afterAll()
+conf.setConf(SQLConf.MAX_TO_STRING_FIELDS, originalMaxToStringFields)
+ }
+
+ private val regenerateGoldenFiles: Boolean =
System.getenv("SPARK_GENERATE_GOLDEN_FILES") == "1"
+
+ protected val baseResourcePath = {
+// use the same way as `SQLQueryTestSuite` to get the resource path
+java.nio.file.Paths.get("src", "test", "resources",
"tpcds-plan-stability").toFile
+ }
+
+ def goldenFilePath: String
+
+ private def getDirForTest(name: String): File = {
+new File(goldenFilePath, name)
+ }
+
+ private def isApproved(dir: File, actualSimplifiedPlan: String): Boolean = {
+val file = new File(dir, "simplified.txt")
+val approved = FileUtils.readFileToString(file, StandardCharsets.UTF_8)
+approved == actualSimplifiedPlan
+ }
+
+ /**
+ * Serialize and save this SparkPlan.
+ * The resulting file is used by [[checkWithApproved]] to check stability.
+ *
+ * @param planthe SparkPlan
+ * @param namethe name of the query
+ * @param explain the full explain output; this is saved to help debug later
as the simplified
+ *plan is not too useful for debugging
+ */
+ private def generateApprovedPlanFile(plan: SparkPlan, name: String, explain:
String): Unit = {
+val dir = getDirForTest(name)
+val simplified = getSimplifiedPlan(plan)
+val foundMatch = dir.exists() && isApproved(dir, simplified)
+
+if (!foundMatch) {
+ FileUtils.deleteDirectory(dir)
+ assert(dir.mkdirs())
+
+ val file = new File(dir, "simplified.txt")
+ FileUtils.writeStringToFile(file, simplified,
[GitHub] [spark] Ngone51 commented on a change in pull request #29270: [SPARK-32466][TEST][SQL] Add PlanStabilitySuite to detect SparkPlan regression
Ngone51 commented on a change in pull request #29270:
URL: https://github.com/apache/spark/pull/29270#discussion_r469653467
##
File path: sql/core/src/test/scala/org/apache/spark/sql/PlanStabilitySuite.scala
##
@@ -0,0 +1,316 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ *http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.sql
+
+import java.io.File
+import java.nio.charset.StandardCharsets
+
+import scala.collection.mutable
+
+import org.apache.commons.io.FileUtils
+
+import org.apache.spark.sql.catalyst.expressions.AttributeSet
+import org.apache.spark.sql.catalyst.util._
+import org.apache.spark.sql.execution._
+import org.apache.spark.sql.execution.adaptive.DisableAdaptiveExecutionSuite
+import org.apache.spark.sql.execution.exchange.{Exchange, ReusedExchangeExec}
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * Check that TPCDS SparkPlans don't change.
+ * If there are plan differences, the error message looks like this:
+ * Plans did not match:
+ * last approved simplified plan:
../tpcds-plan-stability/approved-plans-xxx/q1/simplified.txt
+ * last approved explain plan:
../tpcds-plan-stability/approved-plans-xxx/q1/explain.txt
+ * [last approved simplified plan]
+ *
+ * actual simplified plan: /path/to/tmp/q1.actual.simplified.txt
+ * actual explain plan: /path/to/tmp/q1.actual.explain.txt
+ * [actual simplified plan]
+ * The explain files are saved to help debug later, they are not checked. Only
the simplified
+ * plans are checked (by string comparison).
+ *
+ * Approving new plans:
+ * IF the plan change is intended then re-running the test
+ * with environ var SPARK_GENERATE_GOLDEN_FILES=1 will make the new plan canon.
+ * This should be done only for the queries that need it, to avoid unnecessary
diffs in the
+ * other approved plans.
+ * This can be done by running sbt test-only *PlanStability[WithStats]Suite*
-- -z "q31"
+ * The new plan files should be part of the PR and reviewed.
+ */
+trait PlanStabilitySuite extends TPCDSBase with DisableAdaptiveExecutionSuite {
+
+ private val originalMaxToStringFields = conf.maxToStringFields
+
+ override def beforeAll(): Unit = {
+conf.setConf(SQLConf.MAX_TO_STRING_FIELDS, Int.MaxValue)
+super.beforeAll()
+ }
+
+ override def afterAll(): Unit = {
+super.afterAll()
+conf.setConf(SQLConf.MAX_TO_STRING_FIELDS, originalMaxToStringFields)
+ }
+
+ private val regenerateGoldenFiles: Boolean =
System.getenv("SPARK_GENERATE_GOLDEN_FILES") == "1"
+
+ protected val baseResourcePath = {
+// use the same way as `SQLQueryTestSuite` to get the resource path
+java.nio.file.Paths.get("src", "test", "resources",
"tpcds-plan-stability").toFile
+ }
+
+ def goldenFilePath: String
+
+ private def getDirForTest(name: String): File = {
+new File(goldenFilePath, name)
+ }
+
+ private def isApproved(name: String, dir: File, actualSimplifiedPlan:
String): Boolean = {
+val file = new File(dir, "simplified.txt")
+val approved = FileUtils.readFileToString(file, StandardCharsets.UTF_8)
+approved == actualSimplifiedPlan
+ }
+
+ /**
+ * Serialize and save this SparkPlan.
+ * The resulting file is used by [[checkWithApproved]] to check stability.
+ *
+ * @param planth SparkPlan
+ * @param namethe name of the query
+ * @param explain the full explain output; this is saved to help debug later
as the simplified
+ *plan is not too useful for debugging
+ */
+ private def generateApprovedPlanFile(
+ plan: SparkPlan,
+ name: String,
+ explain: String): Unit = {
+val dir = getDirForTest(name)
+val simplified = getSimplifiedPlan(plan)
+val foundMatch = dir.exists() && isApproved(name, dir, simplified)
+
+if (!foundMatch) {
+ FileUtils.deleteDirectory(dir)
+ assert(dir.mkdirs())
+
+ val file = new File(dir, "simplified.txt")
+ FileUtils.writeStringToFile(file, simplified, StandardCharsets.UTF_8)
+ val fileOriginalPlan = new File(dir, "explain.txt")
+ FileUtils.writeStringToFile(fileOriginalPlan, explain,
StandardCharsets.UTF_8)
+ logInfo(s"APPROVED: $file $fileOriginalPlan")
+}
+ }
+
+ private def checkWithApproved(plan: SparkPlan, name: String,
[GitHub] [spark] Ngone51 commented on a change in pull request #29270: [SPARK-32466][TEST][SQL] Add PlanStabilitySuite to detect SparkPlan regression
Ngone51 commented on a change in pull request #29270:
URL: https://github.com/apache/spark/pull/29270#discussion_r468969318
##
File path: sql/core/src/test/scala/org/apache/spark/sql/PlanStabilitySuite.scala
##
@@ -0,0 +1,312 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ *http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.sql
+
+import java.io.File
+import java.nio.charset.StandardCharsets
+
+import scala.collection.mutable
+
+import org.apache.commons.io.FileUtils
+
+import org.apache.spark.sql.catalyst.expressions.AttributeSet
+import org.apache.spark.sql.catalyst.util._
+import org.apache.spark.sql.execution._
+import org.apache.spark.sql.execution.adaptive.DisableAdaptiveExecutionSuite
+import org.apache.spark.sql.execution.exchange.{Exchange, ReusedExchangeExec}
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * Check that TPCDS SparkPlans don't change.
+ * If there are plan differences, the error message looks like this:
+ * Plans did not match:
+ * last approved plan:
/path/to/tpcds-plan-stability/approved-plans-xxx/q1/simplified.txt
+ * last explain:
/path/to/tpcds-plan-stability/approved-plans-xxx/q1/explain.txt
+ * actual plan: /path/to/tmp/q1.actual.simplified.txt
+ * actual explain: /path/to/tmp/q1.actual.explain.txt
+ * [side by side plan diff]
+ * The explain files are saved to help debug later, they are not checked. Only
the simplified
+ * plans are checked (by string comparison).
+ *
+ * Approving new plans:
+ * IF the plan change is intended then re-running the test
+ * with environ var SPARK_GENERATE_GOLDEN_FILES=1 will make the new plan canon.
+ * This should be done only for the queries that need it, to avoid unnecessary
diffs in the
+ * other approved plans.
+ * This can be done by running sbt test-only *PlanStability[WithStats]Suite*
-- -z "q31"
+ * The new plan files should be part of the PR and reviewed.
+ */
+trait PlanStabilitySuite extends TPCDSBase with DisableAdaptiveExecutionSuite {
Review comment:
This test can fail when enable AQE since AQE could generate different
plans. So, `DisableAdaptiveExecutionSuite` is used to prevent test failure
after enabling AQE.
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] Ngone51 commented on a change in pull request #29270: [SPARK-32466][TEST][SQL] Add PlanStabilitySuite to detect SparkPlan regression
Ngone51 commented on a change in pull request #29270:
URL: https://github.com/apache/spark/pull/29270#discussion_r468968692
##
File path: sql/core/src/test/scala/org/apache/spark/sql/PlanStabilitySuite.scala
##
@@ -0,0 +1,312 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ *http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.sql
+
+import java.io.File
+import java.nio.charset.StandardCharsets
+
+import scala.collection.mutable
+
+import org.apache.commons.io.FileUtils
+
+import org.apache.spark.sql.catalyst.expressions.AttributeSet
+import org.apache.spark.sql.catalyst.util._
+import org.apache.spark.sql.execution._
+import org.apache.spark.sql.execution.adaptive.DisableAdaptiveExecutionSuite
+import org.apache.spark.sql.execution.exchange.{Exchange, ReusedExchangeExec}
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * Check that TPCDS SparkPlans don't change.
+ * If there are plan differences, the error message looks like this:
+ * Plans did not match:
+ * last approved plan:
/path/to/tpcds-plan-stability/approved-plans-xxx/q1/simplified.txt
+ * last explain:
/path/to/tpcds-plan-stability/approved-plans-xxx/q1/explain.txt
+ * actual plan: /path/to/tmp/q1.actual.simplified.txt
+ * actual explain: /path/to/tmp/q1.actual.explain.txt
+ * [side by side plan diff]
+ * The explain files are saved to help debug later, they are not checked. Only
the simplified
+ * plans are checked (by string comparison).
+ *
+ * Approving new plans:
+ * IF the plan change is intended then re-running the test
+ * with environ var SPARK_GENERATE_GOLDEN_FILES=1 will make the new plan canon.
+ * This should be done only for the queries that need it, to avoid unnecessary
diffs in the
+ * other approved plans.
+ * This can be done by running sbt test-only *PlanStability[WithStats]Suite*
-- -z "q31"
+ * The new plan files should be part of the PR and reviewed.
+ */
+trait PlanStabilitySuite extends TPCDSBase with DisableAdaptiveExecutionSuite {
+
+ private val originalMaxToStringFields = conf.maxToStringFields
+
+ override def beforeAll(): Unit = {
+conf.setConf(SQLConf.MAX_TO_STRING_FIELDS, 100)
Review comment:
make sense.
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] Ngone51 commented on a change in pull request #29270: [SPARK-32466][TEST][SQL] Add PlanStabilitySuite to detect SparkPlan regression
Ngone51 commented on a change in pull request #29270:
URL: https://github.com/apache/spark/pull/29270#discussion_r464305243
##
File path: sql/core/src/test/scala/org/apache/spark/sql/PlanStabilitySuite.scala
##
@@ -0,0 +1,300 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ *http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.sql
+
+import java.io.File
+import java.nio.charset.StandardCharsets
+
+import scala.collection.mutable
+
+import org.apache.commons.io.FileUtils
+
+import org.apache.spark.sql.catalyst.expressions.AttributeSet
+import org.apache.spark.sql.catalyst.util._
+import org.apache.spark.sql.execution._
+import org.apache.spark.sql.execution.adaptive.DisableAdaptiveExecutionSuite
+import org.apache.spark.sql.execution.exchange.{Exchange, ReusedExchangeExec}
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * Check that TPCDS SparkPlans don't change.
+ * If there are plan differences, the error message looks like this:
+ * Plans did not match:
+ * last approved plan:
/path/to/tpcds-plan-stability/approved-plans-xxx/q1/0.simplified.txt
+ * last explain:
/path/to/tpcds-plan-stability/approved-plans-xxx/q1/0.explain.txt
+ * actual plan: /path/to/tmp/q1.actual.simplified.txt
+ * actual explain: /path/to/tmp/q1.actual.explain.txt
+ * [side by side plan diff]
+ * The explain files are saved to help debug later, they are not checked. Only
the simplified
+ * plans are checked (by string comparison).
+ *
+ * Approving new plans:
+ * IF the plan change is intended then re-running the test
+ * with environ var SPARK_GENERATE_GOLDEN_FILES=1 will make the new plan canon.
+ * This should be done only for the queries that need it, to avoid unnecessary
diffs in the
+ * other approved plans.
+ * This can be done by running sbt test-only *PlanStabilitySuite* -- -z "(q31)"
+ * The new plan files should be part of the PR and reviewed.
+ *
+ * Multiple approved plans:
+ * It's possible that a query has multiple correct plans. This should be
decided as part of the
+ * review. In this case, change the call to approvePlans and set variant=true.
Review comment:
I was trying to be consistent with Spark's original assumption that a
query may have multiple physical plans and Spark needs to select the best one
bases on some rules, e.g. cost-based rules. But it seems this gets unnecessary
complexity since Spark doesn't really support that yet.
And now, @cloud-fan @maropu have another idea which is to use `variant` to
improve test coverage for the plan. e.g., we could test SMJ as well as BHJ for
a certain plan basing on different table stats.
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
-
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
[GitHub] [spark] Ngone51 commented on a change in pull request #29270: [SPARK-32466][TEST][SQL] Add PlanStabilitySuite to detect SparkPlan regression
Ngone51 commented on a change in pull request #29270:
URL: https://github.com/apache/spark/pull/29270#discussion_r464300780
##
File path: sql/core/src/test/scala/org/apache/spark/sql/PlanStabilitySuite.scala
##
@@ -0,0 +1,300 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ *http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.sql
+
+import java.io.File
+import java.nio.charset.StandardCharsets
+
+import scala.collection.mutable
+
+import org.apache.commons.io.FileUtils
+
+import org.apache.spark.sql.catalyst.expressions.AttributeSet
+import org.apache.spark.sql.catalyst.util._
+import org.apache.spark.sql.execution._
+import org.apache.spark.sql.execution.adaptive.DisableAdaptiveExecutionSuite
+import org.apache.spark.sql.execution.exchange.{Exchange, ReusedExchangeExec}
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * Check that TPCDS SparkPlans don't change.
+ * If there are plan differences, the error message looks like this:
+ * Plans did not match:
+ * last approved plan:
/path/to/tpcds-plan-stability/approved-plans-xxx/q1/0.simplified.txt
+ * last explain:
/path/to/tpcds-plan-stability/approved-plans-xxx/q1/0.explain.txt
+ * actual plan: /path/to/tmp/q1.actual.simplified.txt
+ * actual explain: /path/to/tmp/q1.actual.explain.txt
+ * [side by side plan diff]
+ * The explain files are saved to help debug later, they are not checked. Only
the simplified
+ * plans are checked (by string comparison).
+ *
+ * Approving new plans:
+ * IF the plan change is intended then re-running the test
+ * with environ var SPARK_GENERATE_GOLDEN_FILES=1 will make the new plan canon.
+ * This should be done only for the queries that need it, to avoid unnecessary
diffs in the
+ * other approved plans.
+ * This can be done by running sbt test-only *PlanStabilitySuite* -- -z "(q31)"
+ * The new plan files should be part of the PR and reviewed.
+ *
+ * Multiple approved plans:
+ * It's possible that a query has multiple correct plans. This should be
decided as part of the
+ * review. In this case, change the call to approvePlans and set variant=true.
+ */
+trait PlanStabilitySuite extends TPCDSBase with DisableAdaptiveExecutionSuite {
+
+ private val originalMaxToStringFields = conf.maxToStringFields
+
+ override def beforeAll(): Unit = {
+conf.setConf(SQLConf.MAX_TO_STRING_FIELDS, 100)
+super.beforeAll()
+
+ }
+
+ override def afterAll(): Unit = {
+super.afterAll()
+conf.setConf(SQLConf.MAX_TO_STRING_FIELDS, originalMaxToStringFields)
+ }
+
+ private val regenerateGoldenFiles: Boolean =
System.getenv("SPARK_GENERATE_GOLDEN_FILES") == "1"
+
+ protected val baseResourcePath = {
+// use the same way as `SQLQueryTestSuite` to get the resource path
+java.nio.file.Paths.get("src", "test", "resources",
"tpcds-plan-stability").toFile
+ }
+
+ def goldenFilePath: String
+
+ private def getExistingVariants(name: String): Array[Int] = {
+val dir = new File(goldenFilePath, name)
+// file paths are the form ../q3/2.simplified.txt
+val rgx = """(\d+).simplified.txt""".r
+
+dir.listFiles().filter(_.getName.contains("simplified")).map { file =>
+ val rgx(numStr) = file.getName
+ numStr.toInt
+}
+ }
+
+ private def getNextVariantNumber(name: String): Int = {
+val existingVariants = getExistingVariants(name)
+if (existingVariants.isEmpty) {
+ 0
+} else {
+ existingVariants.max + 1
+}
+ }
+
+ private def getDirForTest(name: String): File = {
+new File(goldenFilePath, name)
+ }
+
+ private def isApproved(name: String, dir: File, actualSimplifiedPlan:
String): Boolean = {
+val existingVariants = getExistingVariants(name)
+existingVariants.exists { variant =>
+ val file = new File(dir, s"$variant.simplified.txt")
+ val approved = FileUtils.readFileToString(file, StandardCharsets.UTF_8)
+ approved == actualSimplifiedPlan
+}
+ }
+
+ /**
+ * Serialize and save this SparkPlan.
+ * The resulting file is used by [[checkWithApproved]] to check stability.
+ *
+ * @param planthe [[SparkPlan]]
+ * @param namethe name of the query
+ * @param variant if false, this plan will become the only approved plan,
otherwise
+ *
[GitHub] [spark] Ngone51 commented on a change in pull request #29270: [SPARK-32466][TEST][SQL] Add PlanStabilitySuite to detect SparkPlan regression
Ngone51 commented on a change in pull request #29270:
URL: https://github.com/apache/spark/pull/29270#discussion_r462769211
##
File path: sql/core/src/test/scala/org/apache/spark/sql/PlanStabilitySuite.scala
##
@@ -0,0 +1,306 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ *http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.sql
+
+import java.io.File
+import java.nio.charset.{Charset, StandardCharsets}
+
+import scala.collection.mutable
+
+import org.apache.commons.io.FileUtils
+
+import org.apache.spark.sql.catalyst.expressions.AttributeSet
+import org.apache.spark.sql.catalyst.util._
+import org.apache.spark.sql.execution._
+import org.apache.spark.sql.execution.adaptive.DisableAdaptiveExecutionSuite
+import org.apache.spark.sql.execution.exchange.{Exchange, ReusedExchangeExec}
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * Check that TPCDS SparkPlans don't change.
+ * If there is a regression, the error message looks like this:
Review comment:
oh yeah. reworded.
##
File path: sql/core/src/test/scala/org/apache/spark/sql/PlanStabilitySuite.scala
##
@@ -0,0 +1,306 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ *http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.spark.sql
+
+import java.io.File
+import java.nio.charset.{Charset, StandardCharsets}
+
+import scala.collection.mutable
+
+import org.apache.commons.io.FileUtils
+
+import org.apache.spark.sql.catalyst.expressions.AttributeSet
+import org.apache.spark.sql.catalyst.util._
+import org.apache.spark.sql.execution._
+import org.apache.spark.sql.execution.adaptive.DisableAdaptiveExecutionSuite
+import org.apache.spark.sql.execution.exchange.{Exchange, ReusedExchangeExec}
+import org.apache.spark.sql.internal.SQLConf
+
+/**
+ * Check that TPCDS SparkPlans don't change.
+ * If there is a regression, the error message looks like this:
+ * Plans did not match:
+ * last approved plan:
/path/to/tpcds-plan-stability/approved-plans-xxx/q1/0.simplified.txt
+ * last explain:
/path/to/tpcds-plan-stability/approved-plans-xxx/q1/0.explain.txt
+ * actual plan: /path/to/tmp/q1.actual.simplified.txt
+ * actual explain: /path/to/tmp/q1.actual.explain.txt
+ * [side by side plan diff]
+ * The explain files are saved to help debug later, they are not checked. Only
the simplified
+ * plans are checked (by string comparison).
+ *
+ * Approving new plans:
+ * IF the plan change is intended then re-running the test
+ * with environ var SPARK_GENERATE_GOLDEN_FILES=1 will make the new plan canon.
+ * This should be done only for the queries that need it, to avoid unnecessary
diffs in the
+ * other approved plans.
+ * This can be done by running sbt test-only *PlanStabilitySuite* -- -z "(q31)"
+ * The new plan files should be part of the PR and reviewed.
+ *
+ * Multiple approved plans:
+ * It's possible that a query has multiple correct plans. This should be
decided as part of the
+ * review. In this case, change the call to approvePlans and set variant=true.
+ */
+trait PlanStabilitySuite extends TPCDSBase with DisableAdaptiveExecutionSuite {
+
+ private var originalMaxToStringFields = conf.maxToStringFields
Review comment:
changed to `val`
##
File path: sql/core/src/test/scala/org/apache/spark/sql/PlanStabilitySuite.scala
##
@@ -0,0 +1,306 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional
