[ 
https://issues.apache.org/jira/browse/HADOOP-957?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Sameer Paranjpye updated HADOOP-957:
------------------------------------

    Component/s: mapred
    Description: 
When creating a map/reduce job with SequenceFileFormat, 
the user would like to add some metada to the output files automatically. 
In particular, if the output value class was a JuteRecord class generated from 
a Jute IDL, 
we would like to add JUTE_IDL/IDL_STRING as a attribute/value pair of the 
metadata.
This way, the output files will be self describing: 
When an application that tries to use the files may not have the value class 
with it. 
But the application can use Jute tool to generate the classes on demand.
Or better yet, the SequenceFile record reader may be able to do that 
automatically.




  was:

When creating a map/reduce job with SequenceFileFormat, 
the user would like to add some metada to the output files automatically. 
In particular, if the output value class was a JuteRecord class generated from 
a Jute IDL, 
we would like to add JUTE_IDL/IDL_STRING as a attribute/value pair of the 
metadata.
This way, the output files will be self describing: 
When an application that tries to use the files may not have the value class 
with it. 
But the application can use Jute tool to generate the classes on demand.
Or better yet, the SequenceFile record reader may be able to do that 
automatically.





> Map/Reduce job with SequenceFileOutputFormat should be able to add user 
> specified metadata to the output file
> -------------------------------------------------------------------------------------------------------------
>
>                 Key: HADOOP-957
>                 URL: https://issues.apache.org/jira/browse/HADOOP-957
>             Project: Hadoop
>          Issue Type: New Feature
>          Components: mapred
>            Reporter: Runping Qi
>
> When creating a map/reduce job with SequenceFileFormat, 
> the user would like to add some metada to the output files automatically. 
> In particular, if the output value class was a JuteRecord class generated 
> from a Jute IDL, 
> we would like to add JUTE_IDL/IDL_STRING as a attribute/value pair of the 
> metadata.
> This way, the output files will be self describing: 
> When an application that tries to use the files may not have the value class 
> with it. 
> But the application can use Jute tool to generate the classes on demand.
> Or better yet, the SequenceFile record reader may be able to do that 
> automatically.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.

Reply via email to