You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-issues@hadoop.apache.org by "Raghu Angadi (JIRA)" <ji...@apache.org> on 2009/09/22 20:48:17 UTC
[jira] Commented: (HADOOP-6218) Split TFile by Record Sequence
Number
[ https://issues.apache.org/jira/browse/HADOOP-6218?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12758355#action_12758355 ]
Raghu Angadi commented on HADOOP-6218:
--------------------------------------
Preliminary review:
The patch looks good.
The patch changes the 'createScanner' API names. For 0.20.2 and 0.21 can we keep the old constructors as well (deprecated in 0.21). I think this is required for 0.20.x at least.
Couple new public methods need JavaDoc.
We need add a unit test.
This is not a ondisk format change for TFile. Mainly exposes the record count it already kept. In that sense I think it is safe for 0.20 and 0.21.
> Split TFile by Record Sequence Number
> -------------------------------------
>
> Key: HADOOP-6218
> URL: https://issues.apache.org/jira/browse/HADOOP-6218
> Project: Hadoop Common
> Issue Type: New Feature
> Reporter: Hong Tang
> Attachments: hadoop-6218-20090827.patch
>
>
> It would be nice if TFile can be split by Record Sequence Number. This way, columnar storage like PIG-833 can align fields that belong to the same row but in different columns.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.