[
https://issues.apache.org/jira/browse/HADOOP-1785?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12523046
]
stack commented on HADOOP-1785:
-------------------------------
Thanks for the patch Ning (and for including a test for your fix).
The patch is missing TestMultiRegionTableMapReduce. Lines should be < 80
characters if possible. Usually folks make their patch inside the
HADOOP_TRUNK dir. The paths in your patch are all absolute paths starting from
C:/).
Otherwise the patch looks great (Making addContent static is a good idea).
Should we refactor splitTable to use your new MultiRegionTable.java?
> [HBase] TableInputFormat.TableRecordReader.next has a bug
> ---------------------------------------------------------
>
> Key: HADOOP-1785
> URL: https://issues.apache.org/jira/browse/HADOOP-1785
> Project: Hadoop
> Issue Type: Bug
> Components: contrib/hbase
> Reporter: Ning Li
> Priority: Minor
> Attachments: TableInputFormat.patch
>
>
> TableInputFormat.TableRecordReader.next has the following:
> if (m_endRow.getLength() > 0 && (tKey.getRow().compareTo(m_endRow) < 0))
> hasMore = false;
> It should be:
> if (m_endRow.getLength() > 0 && (tKey.getRow().compareTo(m_endRow) >= 0))
> hasMore = false;
> TestTableMapReduce test passes because it tested map-reduce on a
> single-region table.
> The problem only shows on a multi-region table. The patch includes changes to
> TestTableMapReduce
> so map-reduce is tested on a multi-region table as well. The multi-region
> table is created
> in a similar way as done in TestSplit.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
