Ryan Williams created SPARK-11461:
-------------------------------------
Summary: ObjectFile saving/loading should use configured serializer
Key: SPARK-11461
URL: https://issues.apache.org/jira/browse/SPARK-11461
Project: Spark
Issue Type: Improvement
Components: Spark Core
Affects Versions: 1.5.1
Reporter: Ryan Williams
[RDD.saveAsObjectFile|https://github.com/apache/spark/blob/v1.5.1/core/src/main/scala/org/apache/spark/rdd/RDD.scala#L1452]
and
[SparkContext.objectFile|https://github.com/apache/spark/blob/v1.5.1/core/src/main/scala/org/apache/spark/SparkContext.scala#L1223]
use
[Utils.serialize|https://github.com/apache/spark/blob/v1.5.1/core/src/main/scala/org/apache/spark/util/Utils.scala#L78-L85]
and
[Utils.deserialize|https://github.com/apache/spark/blob/v1.5.1/core/src/main/scala/org/apache/spark/util/Utils.scala#L94-L105]
which are hard-coded to use Java SerDe rather than the serializer configured
via the {{spark.serializer}} conf param.
I'd like to write RDDs as Object-/Sequence-Files using e.g. Kryo serde instead
of Java; is there a way to do this, or any reason that Spark currently only
supports Java?
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
