[
https://issues.apache.org/jira/browse/FLINK-4771?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15562132#comment-15562132
]
ASF GitHub Bot commented on FLINK-4771:
---------------------------------------
Github user fhueske commented on a diff in the pull request:
https://github.com/apache/flink/pull/2612#discussion_r82591316
--- Diff:
flink-batch-connectors/flink-avro/src/main/java/org/apache/flink/api/java/io/AvroOutputFormat.java
---
@@ -36,6 +38,8 @@
private final Class<E> avroValueType;
private transient Schema userDefinedSchema = null;
+
+ private transient CodecFactory codecFactory = null;
--- End diff --
Flink uses Java Serialization to ship code from the client to the cluster.
`CodecFactory` does not implement `Serializable`. Therefore, the `codecFactory`
field must be declared as `transient`. However, the `writeObject` and
`readObject` methods must be adapted to manually serialize / deserialize the
information. Otherwise, the `CodecFactory` is lost when the object is
deserialized at the client.
> Compression for AvroOutputFormat
> --------------------------------
>
> Key: FLINK-4771
> URL: https://issues.apache.org/jira/browse/FLINK-4771
> Project: Flink
> Issue Type: Improvement
> Components: Batch Connectors and Input/Output Formats
> Reporter: Lars Bachmann
> Original Estimate: 1h
> Remaining Estimate: 1h
>
> Currently it is not possible to set a compression codec for the
> AvroOutputFormat.
> This improvement will provide a setter for the avro CodecFactory which is
> used by the DataFileWriter.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
