[
https://issues.apache.org/jira/browse/HADOOP-4199?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12633086#action_12633086
]
Pete Wyckoff commented on HADOOP-4199:
--------------------------------------
I guess this JIRA brings up the larger point that although a
Serializer/Deserializer may be on something that extends Writable, you may
still need more information that that? i.e., Binary, CSV, .. for RecordIO.
This isn't a great example because admittedly CSV isn't a very useful format
for long lived data.
So, maybe I should mark this as invalid?
But, what if i did have data in CSV format? I could never get to it with the
current Serialization framework and SequenceFileRecordReader - I would actually
have to define my own SequenceFileRecordReader that knows not to add the
default Writable Serialization implementation??
> Add Serialization for RecordIO
> ------------------------------
>
> Key: HADOOP-4199
> URL: https://issues.apache.org/jira/browse/HADOOP-4199
> Project: Hadoop Core
> Issue Type: New Feature
> Components: contrib/serialization, mapred
> Reporter: Pete Wyckoff
> Priority: Minor
> Attachments: HADOOP-4199.0.txt, RecordIOSerialization.java
>
>
> Implement
> org.apache.hadoop.io.serialization.Serialization/Serializer/Deserializer
> interfaces
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
