1u0 commented on a change in pull request #8467: [FLINK-12535][network] Make CheckpointBarrierHandler non-blocking URL: https://github.com/apache/flink/pull/8467#discussion_r287316831
########## File path: flink-runtime/src/main/java/org/apache/flink/runtime/io/AsyncDataInput.java ########## @@ -0,0 +1,84 @@ +/* + * 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.flink.runtime.io; + +import org.apache.flink.annotation.Internal; + +import java.util.concurrent.CompletableFuture; + +/** + * Interface defining couple of essential methods for asynchronous and non blocking data polling. + * + * <p>For the most efficient usage, user of this class is suppose to call {@link #pollNext()} + * until it returns that no more elements are available. If that happens, he should check if + * input {@link #isFinished()}. If not, he should wait for {@link #isAvailable()} + * {@link CompletableFuture} to be completed. For example: + * + * <pre> + * {@code + * while (!input.isFinished()) { + * AsyncDataInput<Optional<T>> input = ...; + * Optional<T> next; + * + * while (true) { + * next = input.pollNext(); + * if (!next.isPresent()) { + * break; + * } + * // do something with next + * } + * + * input.isAvailable().get(); + * } + * } + * </pre> + */ +@Internal +public interface AsyncDataInput<T> { + CompletableFuture<?> AVAILABLE = CompletableFuture.completedFuture(null); + + /** + * Poll the next element. + * + * @return + * <ol> + * <li> + * If the {@link T} is an {@code Optional}: return {@code Optional.of(element)} - + * next returned element. {@code Optional.empty()} will be returned if there is no data + * to return or if {@link #isFinished()} returns true. + * </li> + * <li> + * If next element or {@code null} if there is no data to return or if + * {@link #isFinished()} returns true. + * </li> + * </ol> + */ + T pollNext() throws Exception; Review comment: The `Input` interface in #8476 PR uses `AsyncDataInput<StreamElement>` and uses `null` as sentinel (the interface doesn't use `Optional<StreamElement>` due to performance concerns). We have discussed offline with @pnowojski and there were some alternative ideas how the interface of `AsyncDataInput` may be more explicit (so the contract of `pollNext()` method is not overloaded by `Optional.empty()` or `null` return value as marker) and more cleaner (at least, from my point of view). I'll just leave one of the ideas that I find fine (but not enforced). The `AsyncDataInput<T>` could have been extended to have two `poll*` methods, with the same behavior regarding traversal of input, but with just different caller conventions. So it's up to the caller to decide which flavor they prefer: ```java interface AsyncDataInput<T> { ... Optional<T> pollNext(); @Nullable default T pollNextOrNull() { return pollNext().orElse(null); } } ``` ---------------------------------------------------------------- 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 With regards, Apache Git Services
