Copilot commented on code in PR #12444: URL: https://github.com/apache/gluten/pull/12444#discussion_r3549897024
########## backends-velox/src-delta40/test/scala/org/apache/spark/sql/delta/GlutenDeltaVariantWriteSuite.scala: ########## @@ -0,0 +1,91 @@ +/* + * 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.delta + +import org.apache.spark.sql.QueryTest +import org.apache.spark.sql.Row +import org.apache.spark.sql.delta.test.DeltaSQLCommandTest +import org.apache.spark.sql.test.SharedSparkSession + +/** + * Velox has no Arrow representation for VariantType, so a Delta write that Gluten offloads to the + * native columnar writer (DataFrameWriter.save, UPDATE, ...) used to throw + * `UnsupportedOperationException: Unsupported data type: variant` at runtime. + * GlutenOptimisticTransaction now detects a schema Velox cannot write (such as one containing a + * variant) and delegates such writes to the vanilla Delta write path. These tests use offloaded + * writes (save and UPDATE) so they exercise that path; with INSERT INTO the write is not offloaded + * and would pass regardless of the fix. Review Comment: The suite-level scaladoc says these are "offloaded" writes, but after this change the whole point is that these writes *would* be offloaded when native write is enabled and must now fall back to vanilla Delta because the schema is unsupported. The current wording is misleading about what the tests are exercising. ########## backends-velox/src-delta40/test/scala/org/apache/spark/sql/delta/GlutenDeltaVariantWriteSuite.scala: ########## @@ -0,0 +1,91 @@ +/* + * 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.delta + +import org.apache.spark.sql.QueryTest +import org.apache.spark.sql.Row +import org.apache.spark.sql.delta.test.DeltaSQLCommandTest +import org.apache.spark.sql.test.SharedSparkSession + +/** + * Velox has no Arrow representation for VariantType, so a Delta write that Gluten offloads to the + * native columnar writer (DataFrameWriter.save, UPDATE, ...) used to throw + * `UnsupportedOperationException: Unsupported data type: variant` at runtime. + * GlutenOptimisticTransaction now detects a schema Velox cannot write (such as one containing a + * variant) and delegates such writes to the vanilla Delta write path. These tests use offloaded + * writes (save and UPDATE) so they exercise that path; with INSERT INTO the write is not offloaded + * and would pass regardless of the fix. + */ +class GlutenDeltaVariantWriteSuite + extends QueryTest + with SharedSparkSession + with DeltaSQLCommandTest { Review Comment: `DeltaSQLCommandTest` already extends `SharedSparkSession`, so mixing in `with SharedSparkSession` again is redundant and can make trait linearization / SparkSession setup harder to reason about. ########## backends-velox/src-delta40/main/scala/org/apache/spark/sql/delta/GlutenOptimisticTransaction.scala: ########## @@ -48,6 +48,15 @@ class GlutenOptimisticTransaction(delegate: OptimisticTransaction) writeOptions: Option[DeltaOptions], isOptimize: Boolean, additionalConstraints: Seq[Constraint]): Seq[FileAction] = { + // Velox cannot write every Spark type: the native write path inserts a RowToVeloxColumnarExec + // transition whose SparkArrowUtil.toArrowSchema call throws (e.g. `Unsupported data type: + // variant`) for any type it has no Arrow mapping for. Reuse the backend's schema validator so + // such writes fall back to the vanilla Delta write path instead of failing at runtime. + val unsupportedReason = VeloxValidatorApi.validateSchema(inputData.schema) + if (unsupportedReason.isDefined) { + logInfo(s"Schema is not supported by Velox (${unsupportedReason.get}); falling back to the vanilla Delta write path.") + return super.writeFiles(inputData, writeOptions, isOptimize, additionalConstraints) + } Review Comment: Avoid calling `Option.get` in log messages; it’s safe here due to the prior `isDefined` check, but using pattern matching makes the intent clearer and avoids repeating `unsupportedReason` lookups. -- 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]
