hadoop-common-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Owen O'Malley (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HADOOP-4057) MultipleFileInputFormat should support a sub InputFormat
Date Wed, 03 Sep 2008 04:39:44 GMT

    [ https://issues.apache.org/jira/browse/HADOOP-4057?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12627899#action_12627899

Owen O'Malley commented on HADOOP-4057:

I'd suggest something like this:

public class MultipleFileInputFormat implements InputFormat {
public static void setUserInputFormat(Configuration conf, Class<? extends InputFormat>
public static Class<? extends InputFormat> getUserInputFormat(Configuration conf);

public InputSplit[] getSplits(JobConf conf, int maps) {
  InputFormat userInputFormat = ReflectionUtils.newInstance(getUserInputFormat(conf), conf);
  InputSplit[] userSplits = userInputFormat.getSPlits(conf, maps);
  ... merge them together ...
  return exactly maps splits

public RecordReader getRecordReader(InputSplit split, JobConf conf, Reporter reporter) {
   .. return a record reader that returns all of the records for split 1, split 2, etc..


> MultipleFileInputFormat should support a sub InputFormat
> --------------------------------------------------------
>                 Key: HADOOP-4057
>                 URL: https://issues.apache.org/jira/browse/HADOOP-4057
>             Project: Hadoop Core
>          Issue Type: Improvement
>          Components: mapred
>            Reporter: Owen O'Malley
> Currently, to use MultipleFileInputFormat you need to create a subclass, but it seems
pretty easy to modify it to use a sub InputFormat and put the results together into bigger

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

View raw message