Mailing-List: contact dev-help@apex.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@apex.apache.org
From: chinmaykolhatkar <git@git.apache.org>
To: dev@apex.incubator.apache.org
Reply-To: dev@apex.incubator.apache.org
References: <git-pr-335-apex-malhar@git.apache.org>
In-Reply-To: <git-pr-335-apex-malhar@git.apache.org>
Subject: [GitHub] apex-malhar pull request #335: [Review Only] APEXMALHAR-1701 Deduper with Ma...
Content-Type: text/plain
Message-Id: <20160801150625.CF635E058E@git1-us-west.apache.org>
Date: Mon,  1 Aug 2016 15:06:25 +0000 (UTC)
archived-at: Mon, 01 Aug 2016 15:06:32 -0000

Github user chinmaykolhatkar commented on a diff in the pull request:

    https://github.com/apache/apex-malhar/pull/335#discussion_r72994134
  
    --- Diff: library/src/main/java/org/apache/apex/malhar/lib/dedup/AbstractDeduper.java ---
    @@ -0,0 +1,462 @@
    +/**
    + * 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.apex.malhar.lib.dedup;
    +
    +import java.util.Iterator;
    +import java.util.Map;
    +import java.util.Map.Entry;
    +import java.util.concurrent.ExecutionException;
    +import java.util.concurrent.Future;
    +
    +import javax.validation.constraints.NotNull;
    +
    +import org.slf4j.Logger;
    +import org.slf4j.LoggerFactory;
    +
    +import org.apache.apex.malhar.lib.state.BucketedState;
    +import org.apache.apex.malhar.lib.state.managed.ManagedTimeUnifiedStateImpl;
    +import org.apache.apex.malhar.lib.state.managed.TimeBucketAssigner;
    +import org.apache.hadoop.classification.InterfaceStability.Evolving;
    +
    +import com.google.common.base.Preconditions;
    +import com.google.common.collect.Maps;
    +import com.google.common.util.concurrent.Futures;
    +
    +import com.datatorrent.api.AutoMetric;
    +import com.datatorrent.api.Context;
    +import com.datatorrent.api.Context.OperatorContext;
    +import com.datatorrent.api.DAG;
    +import com.datatorrent.api.DefaultInputPort;
    +import com.datatorrent.api.DefaultOutputPort;
    +import com.datatorrent.api.Operator;
    +import com.datatorrent.api.Operator.ActivationListener;
    +import com.datatorrent.api.annotation.OperatorAnnotation;
    +import com.datatorrent.api.annotation.OutputPortFieldAnnotation;
    +import com.datatorrent.lib.fileaccess.FileAccessFSImpl;
    +import com.datatorrent.lib.fileaccess.TFileImpl;
    +import com.datatorrent.netlet.util.Slice;
    +
    +/**
    + * Abstract class which allows de-duplicating incoming tuples based on a configured key.
    + * Also supports expiry mechanism based on a configurable expiry period configured using {@link TimeBucketAssigner}
    + * in {@link ManagedTimeUnifiedStateImpl}
    + *
    + * @param <T> type of events
    + */
    +@Evolving
    +@OperatorAnnotation(checkpointableWithinAppWindow = false)
    +public abstract class AbstractDeduper<T>
    +    implements Operator, Operator.IdleTimeHandler, ActivationListener<Context>, Operator.CheckpointNotificationListener
    +{
    +  /**
    +   * The input port on which events are received.
    +   */
    +  public final transient DefaultInputPort<T> input = new DefaultInputPort<T>()
    +  {
    +    @Override
    +    public final void process(T tuple)
    +    {
    +      processTuple(tuple);
    +    }
    +  };
    +
    +  /**
    +   * The output port on which deduped events are emitted.
    +   */
    +  public final transient DefaultOutputPort<T> unique = new DefaultOutputPort<>();
    +
    +  /**
    +   * The output port on which duplicate events are emitted.
    +   */
    +  @OutputPortFieldAnnotation(optional = true)
    +  public final transient DefaultOutputPort<T> duplicate = new DefaultOutputPort<>();
    +
    +  /**
    +   * The output port on which expired events are emitted.
    +   */
    +  @OutputPortFieldAnnotation(optional = true)
    +  public final transient DefaultOutputPort<T> expired = new DefaultOutputPort<>();
    +
    +  /**
    +   * Whether or not the order of tuples be maintained
    +   */
    +  private boolean orderedOutput = false;
    --- End diff --
    
    I suggest we make this true by default and call it out explicitly of performance penalty of doing this.
    The reason for default true is then one can be sure that default deduper is idempotent which usually a assumption while targeting end-to-end exactly once.
    
    Choice should be given on making it performant at the cost of idempotency. Not the other way round.


---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastructure@apache.org or file a JIRA ticket
with INFRA.
---