[
https://issues.apache.org/jira/browse/HADOOP-1986?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12533188
]
Joydeep Sen Sarma commented on HADOOP-1986:
-------------------------------------------
looks great. can we also incorporate Tom's suggestion of an explicit open call
to cache any wrappers that the serializer wants to create around input/output
streams? thrift will need a base class. it allows different serialization
formats - that can be picked up from the configuration. compared to serializing
through byteswritable intermediary - we would save on 4 bytes per serialized
record - as well as less memory copies. not bad.
> Add support for a general serialization mechanism for Map Reduce
> ----------------------------------------------------------------
>
> Key: HADOOP-1986
> URL: https://issues.apache.org/jira/browse/HADOOP-1986
> Project: Hadoop
> Issue Type: New Feature
> Components: mapred
> Reporter: Tom White
> Fix For: 0.16.0
>
> Attachments: SerializableWritable.java
>
>
> Currently Map Reduce programs have to use WritableComparable-Writable
> key-value pairs. While it's possible to write Writable wrappers for other
> serialization frameworks (such as Thrift), this is not very convenient: it
> would be nicer to be able to use arbitrary types directly, without explicit
> wrapping and unwrapping.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
