[GitHub] [spark] attilapiros commented on a change in pull request #31876: [SPARK-34942][API][CORE] Abstract Location in MapStatus to enable support for custom storage

[GitBox <gi...@apache.org>]

attilapiros commented on a change in pull request #31876:
URL: https://github.com/apache/spark/pull/31876#discussion_r807540945



##########
File path: core/src/main/java/org/apache/spark/shuffle/api/Location.java
##########
@@ -0,0 +1,61 @@
+/*
+ * 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 org.apache.spark.annotation.Private;
+
+import java.io.Externalizable;
+import java.io.ObjectInput;
+import java.io.ObjectOutput;
+
+
+/**
+ * :: Private ::
+ * An interface for plugging in the location of shuffle files, in order to support store shuffle
+ * data in different storage, e.g., BlockManager, HDFS, S3. It would be generated by
+ * {@link ShuffleMapOutputWriter} after writing a shuffle data file and used by ShuffleMapOutputReader
+ * to read the shuffle data.
+ *
+ * Since the location is returned by {@link ShuffleMapOutputWriter#commitAllPartitions()} at executor
+ * and would be sent to driver, users must ensure the location is serializable by
+ *
+ *  - implement a 0-arg constructor
+ *  - implement {@link java.io.Externalizable#readExternal(ObjectInput)} for deserialization
+ *  - implement {@link java.io.Externalizable#writeExternal(ObjectOutput)} for serialization
+ *
+ * Since the location will be used as keys in maps or comparing with others, users must ensure that
+ * invoking {@link java.lang.Object#equals(Object)} or {@link java.lang.Object#hashCode()} on the
+ * {@link Location} instances would distinguish the different locations.
+ *
+ * Spark has its own default implementation of {@link Location} as
+ * {@link org.apache.spark.storage.BlockManagerId}, which is a subclass of {@link ExecutorLocation}
+ * since each {@link org.apache.spark.storage.BlockManager} must belong to a certain executor.
+ * And {@link ExecutorLocation} is a subclass of {@link HostLocation} since each executor must
+ * belong to a certain host. Users should choose the appropriate location interface according to their
+ * own use cases.
+ *
+ * :: Caution ::
+ * Spark would reuse the same location instance for locations which are equal due to the
+ * performance concern. Thus, users should also guarantee the implemented {@link Location}
+ * is IMMUTABLE.
+ *
+ * @since 3.2.0
+ */
+@Private
+public interface Location extends Externalizable {

Review comment:
       Technically a single abstract `Location` interface which can be implemented as you like can handle multiple real locations. IMHO it is just naming: Location => Locations and as mostly single real locations are represented I would even keep the current name. 




-- 
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: reviews-unsubscribe@spark.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org