Copilot commented on code in PR #13103: URL: https://github.com/apache/gluten/pull/13103#discussion_r4269051542
########## ep/build-velox/src/get-velox.sh: ########## @@ -25,7 +25,7 @@ RUN_SETUP_SCRIPT=ON ENABLE_ENHANCED_FEATURES=OFF # Developer use only for testing Velox PR. -UPSTREAM_VELOX_PR_ID="" +UPSTREAM_VELOX_PR_ID="19141" Review Comment: This hardcodes the developer-only `UPSTREAM_VELOX_PR_ID` hook to a specific PR, which contradicts the PR description that this hook is temporary and should not be merged. Please revert this to the empty default (or gate it behind an explicit environment variable/flag) so normal builds don’t pin to an upstream PR unexpectedly. ########## gluten-ut/common/src/test/scala/org/apache/spark/sql/GlutenDecimalRoundSuite.scala: ########## @@ -0,0 +1,137 @@ +/* + * 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 org.apache.gluten.config.GlutenConfig +import org.apache.gluten.execution.ProjectExecTransformer + +import org.apache.spark.sql.catalyst.expressions.Round +import org.apache.spark.sql.catalyst.optimizer.NullPropagation +import org.apache.spark.sql.internal.SQLConf +import org.apache.spark.sql.types._ + +import scala.collection.JavaConverters._ + +class GlutenDecimalRoundSuite extends GlutenSQLTestsTrait { + private def withInput(dataType: DecimalType, values: Seq[String])(f: => Unit): Unit = { + val schema = new StructType().add("value", dataType) + val rows = values.map(value => Row(new java.math.BigDecimal(value))) :+ Row(null) + withSQLConf( + SQLConf.ADAPTIVE_EXECUTION_ENABLED.key -> "false", + GlutenConfig.GLUTEN_ANSI_FALLBACK_ENABLED.key -> "false") { + withTempPath { + path => + spark.createDataFrame(rows.asJava, schema).coalesce(1).write.parquet(path.toString) + withTempView("decimal_round_input") { + spark.read.parquet(path.toString).createOrReplaceTempView("decimal_round_input") + f + } + } + } + } + + private def assertExecution(df: DataFrame, native: Boolean): Unit = { + val plan = df.queryExecution.executedPlan + val locations = plan.collect { + case node if node.expressions.exists(_.exists(_.isInstanceOf[Round])) => + node.isInstanceOf[ProjectExecTransformer] + } + assert( + locations.nonEmpty && locations.forall(_ == native), + s"Unexpected ROUND execution:\n$plan") + } + + private def checkRound(scale: String, native: Boolean = true): Unit = { + val query = s"SELECT round(value, $scale) FROM decimal_round_input" + withSQLConf(GlutenConfig.GLUTEN_ENABLED.key -> "false") { + val vanilla = sql(query) + val expectedSchema = vanilla.schema + val expected = vanilla.collect().toSeq + withSQLConf(GlutenConfig.GLUTEN_ENABLED.key -> "true") { + val df = sql(query) + assertExecution(df, native) + assert(df.schema == expectedSchema) + checkAnswer(df, expected) + } + } + } + + testGluten("decimal ROUND preserves schemas and HALF_UP results across storage boundaries") { + val cases = Seq( + (DecimalType(3, 1), Seq("2.5", "-2.5", "0")), + (DecimalType(18, 0), Seq("999999999999999999", "-999999999999999999")), + (DecimalType(19, 0), Seq("9999999999999999999", "-9999999999999999999")), + (DecimalType(38, 1), Seq("9" * 37 + ".9", "-" + "9" * 37 + ".9")), + (DecimalType(38, 38), Seq("0.6", "-0.6", "0.5", "-0.5")) + ) + for ((dataType, values) <- cases) { + withInput(dataType, values) { + for (ansi <- Seq("false", "true")) { + withSQLConf(SQLConf.ANSI_ENABLED.key -> ansi) { + Seq(-400, -39, -1, 0, 1, 38, 400).foreach(scale => checkRound(scale.toString)) + checkRound("-1 + 0") + } + } + } + } + } + + testGluten("decimal ROUND handles the 38-digit HALF_UP tie and precision overflow") { + withInput(DecimalType(38, 0), Seq("4" + "9" * 37, "-4" + "9" * 37)) { + checkRound("-38") + } + for ((value, scale) <- Seq(("9" * 38, -1), ("5" + "0" * 37, -38), ("-5" + "0" * 37, -38))) { + withInput(DecimalType(38, 0), Seq(value)) { + for (ansi <- Seq("false", "true")) { + withSQLConf(SQLConf.ANSI_ENABLED.key -> ansi) { + val query = s"SELECT round(value, $scale) FROM decimal_round_input" + withSQLConf(GlutenConfig.GLUTEN_ENABLED.key -> "false") { + val error = intercept[Exception](sql(query).collect()) + assert(Iterator.iterate[Throwable](error)(_.getCause).takeWhile(_ != null).exists { + cause => + Option(cause.getMessage).exists { + message => + message.contains("NUMERIC_VALUE_OUT_OF_RANGE") || + message.contains("DECIMAL_PRECISION_EXCEEDS_MAX_PRECISION") + } + }) + } + val df = sql(query) + assertExecution(df, native = true) + val error = intercept[Exception](df.collect()) + assert(Iterator.iterate[Throwable](error)(_.getCause).takeWhile(_ != null).exists { + cause => Option(cause.getMessage).exists(_.contains("Decimal overflow")) + }) Review Comment: Asserting a single native error-message substring (`\"Decimal overflow\"`) is brittle across Velox/Gluten versions and may cause avoidable CI failures when diagnostics change. Consider loosening this assertion (e.g., accept a small set of known substrings, or validate via a structured error class/code if available) similar to how the vanilla Spark path allows multiple error identifiers. ########## gluten-ut/common/src/test/scala/org/apache/spark/sql/GlutenDecimalRoundSuite.scala: ########## @@ -0,0 +1,137 @@ +/* + * 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 org.apache.gluten.config.GlutenConfig +import org.apache.gluten.execution.ProjectExecTransformer + +import org.apache.spark.sql.catalyst.expressions.Round +import org.apache.spark.sql.catalyst.optimizer.NullPropagation +import org.apache.spark.sql.internal.SQLConf +import org.apache.spark.sql.types._ + +import scala.collection.JavaConverters._ Review Comment: `scala.collection.JavaConverters._` is deprecated on Scala 2.13+ (used by Spark 4.x). Since this suite is shared across Spark versions, prefer `scala.jdk.CollectionConverters._` (and adjust `.asJava` usage accordingly) to avoid deprecation warnings and reduce forward-compat maintenance costs. -- 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. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
