vanzin commented on a change in pull request #25007: [SPARK-28209][CORE][SHUFFLE] Proposed new shuffle writer API URL: https://github.com/apache/spark/pull/25007#discussion_r302757391
########## File path: core/src/main/java/org/apache/spark/shuffle/api/SupportsTransferTo.java ########## @@ -0,0 +1,53 @@ +/* + * 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.shuffle.api; + +import java.io.IOException; + +import org.apache.spark.annotation.Experimental; + +/** + * :: Experimental :: + * Indicates that partition writers can transfer bytes directly from input byte channels to + * output channels that stream data to the underlying shuffle partition storage medium. + * <p> + * This API is separated out for advanced users because it only needs to be used for + * specific low-level optimizations. The idea is that the returned channel can transfer bytes + * from the input file channel out to the backing storage system without copying data into + * memory. + * <p> + * Most shuffle plugin implementations should use {@link ShufflePartitionWriter} instead. + * + * @since 3.0.0 + */ +@Experimental +public interface SupportsTransferTo extends ShufflePartitionWriter { Review comment: Not a fan of the name. How about making this part of `ShufflePartitionWriter` itself? Have a default implementation for `openTransferrableChannel()` that returns `null`, and if a write supports the fast path, it just needs to override that. I was actually going to suggest only having the method that returns the channel, but there's code that uses streams depending on the value of `spark.file.transferTo`. I think it's worth mentioning that in the javadocs somewhere. ---------------------------------------------------------------- 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: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
