[jira] [Updated] (MRUNIT-193) Serialization.copy throws NPE instead of ISE (missing serilization impl) for Hadoop 2.x

["Cosmin Lehene (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/MRUNIT-193?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Cosmin Lehene updated MRUNIT-193:
---------------------------------

    Description: 
This may be the result of a refactoring. 
The current code attempts to catch a NPE with the intent to detect a missing serialization implementation.

However this behavior differs between Hadoop 1.x and 2.x
Hadoop 1.x will NPE in the first try/catch block, while 2.x will in the second.
{noformat}
    try {
      serializer = (Serializer<Object>) serializationFactory
          .getSerializer(clazz);
      deserializer = (Deserializer<Object>) serializationFactory
          .getDeserializer(clazz);
    } catch (NullPointerException e) {
      throw new IllegalStateException(
          "No applicable class implementing Serialization in conf at io.serializations for "
              + orig.getClass(), e);
    }
    try {
      final DataOutputBuffer outputBuffer = new DataOutputBuffer();
      serializer.open(outputBuffer);
      serializer.serialize(orig);
      final DataInputBuffer inputBuffer = new DataInputBuffer();
      inputBuffer.reset(outputBuffer.getData(), outputBuffer.getLength());
      deserializer.open(inputBuffer);
      return (T) deserializer.deserialize(copy);
    } catch (final IOException e) {
      throw new RuntimeException(e);
    }
{noformat}

Hadoop 1.x 
{code}
  public <T> Serializer<T> getSerializer(Class<T> c) {
    return getSerialization(c).getSerializer(c);
  }
{code}

Hadoop 2.x

{code}
  public <T> Serializer<T> getSerializer(Class<T> c) {
    Serialization<T> serializer = getSerialization(c);
    if (serializer != null) {
      return serializer.getSerializer(c);
    }
    return null;
  }
{code}

  was:
This may be the result of a refactoring. 
The current code attempts to catch a NPE with the intent to detect a missing serialization implementation.

However this behavior differs between Hadoop 1.x and 2.x
Hadoop 1.x will NPE in the first try/catch block, while 2.x will in the second.
{code}
    try {
      serializer = (Serializer<Object>) serializationFactory
          .getSerializer(clazz);
      deserializer = (Deserializer<Object>) serializationFactory
          .getDeserializer(clazz);
    } catch (NullPointerException e) {
      throw new IllegalStateException(
          "No applicable class implementing Serialization in conf at io.serializations for "
              + orig.getClass(), e);
    }
    try {
      final DataOutputBuffer outputBuffer = new DataOutputBuffer();
      serializer.open(outputBuffer);
      serializer.serialize(orig);
      final DataInputBuffer inputBuffer = new DataInputBuffer();
      inputBuffer.reset(outputBuffer.getData(), outputBuffer.getLength());
      deserializer.open(inputBuffer);
      return (T) deserializer.deserialize(copy);
    } catch (final IOException e) {
      throw new RuntimeException(e);
    }
{code}

Hadoop 1.x 
{code}
  public <T> Serializer<T> getSerializer(Class<T> c) {
    return getSerialization(c).getSerializer(c);
  }
{code}

Hadoop 2.x

{code}
  public <T> Serializer<T> getSerializer(Class<T> c) {
    Serialization<T> serializer = getSerialization(c);
    if (serializer != null) {
      return serializer.getSerializer(c);
    }
    return null;
  }
{code}


> Serialization.copy throws NPE instead of ISE (missing serilization impl) for Hadoop 2.x
> ---------------------------------------------------------------------------------------
>
>                 Key: MRUNIT-193
>                 URL: https://issues.apache.org/jira/browse/MRUNIT-193
>             Project: MRUnit
>          Issue Type: Bug
>    Affects Versions: 1.0.0
>            Reporter: Cosmin Lehene
>            Priority: Trivial
>             Fix For: 1.0.0
>
>   Original Estimate: 1h
>  Remaining Estimate: 1h
>
> This may be the result of a refactoring. 
> The current code attempts to catch a NPE with the intent to detect a missing serialization implementation.
> However this behavior differs between Hadoop 1.x and 2.x
> Hadoop 1.x will NPE in the first try/catch block, while 2.x will in the second.
> {noformat}
>     try {
>       serializer = (Serializer<Object>) serializationFactory
>           .getSerializer(clazz);
>       deserializer = (Deserializer<Object>) serializationFactory
>           .getDeserializer(clazz);
>     } catch (NullPointerException e) {
>       throw new IllegalStateException(
>           "No applicable class implementing Serialization in conf at io.serializations for "
>               + orig.getClass(), e);
>     }
>     try {
>       final DataOutputBuffer outputBuffer = new DataOutputBuffer();
>       serializer.open(outputBuffer);
>       serializer.serialize(orig);
>       final DataInputBuffer inputBuffer = new DataInputBuffer();
>       inputBuffer.reset(outputBuffer.getData(), outputBuffer.getLength());
>       deserializer.open(inputBuffer);
>       return (T) deserializer.deserialize(copy);
>     } catch (final IOException e) {
>       throw new RuntimeException(e);
>     }
> {noformat}
> Hadoop 1.x 
> {code}
>   public <T> Serializer<T> getSerializer(Class<T> c) {
>     return getSerialization(c).getSerializer(c);
>   }
> {code}
> Hadoop 2.x
> {code}
>   public <T> Serializer<T> getSerializer(Class<T> c) {
>     Serialization<T> serializer = getSerialization(c);
>     if (serializer != null) {
>       return serializer.getSerializer(c);
>     }
>     return null;
>   }
> {code}



--
This message was sent by Atlassian JIRA
(v6.1#6144)