hadoop-mapreduce-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "E. Sammer (JIRA)" <j...@apache.org>
Subject [jira] Commented: (MAPREDUCE-2226) TaggedInputSplit should be public
Date Tue, 21 Dec 2010 20:07:02 GMT

    [ https://issues.apache.org/jira/browse/MAPREDUCE-2226?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12973855#action_12973855
] 

E. Sammer commented on MAPREDUCE-2226:
--------------------------------------

I can definitely do that if it makes sense to have it be part of this issue. So:

* all methods except for getInputSplit() should be marked @InterfaceAudience.Private.
* TaggedInputSplit javadoc should be updated to indicate it exists for MI functionality.
* MI javadoc should be updated to illustrate the use case of getting at the original InputSplits
when TaggedInputSplit is in use.

Anything else?

> TaggedInputSplit should be public
> ---------------------------------
>
>                 Key: MAPREDUCE-2226
>                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-2226
>             Project: Hadoop Map/Reduce
>          Issue Type: Improvement
>    Affects Versions: 0.20.1
>            Reporter: E. Sammer
>            Assignee: E. Sammer
>         Attachments: MAPREDUCE-2226.diff
>
>
> Currently it's not possible to get at the original InputSplits when using MultipleInputs.
This is because TaggedInputSplit (used by DelegatingInputFormat used by MultipleInputs) is
not public. This means things like the following do not work:
> {code}
> public void map(...) {
>   FileSplit fis = (FileSplit)((TaggedInputSplit) reporter.getInputSplit()).getInputSplit();
>   Path p = fis.getPath();
> }
> {code}
> This prevents users from getting at input split specific data.

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


Mime
View raw message