[
https://issues.apache.org/jira/browse/HIVE-6060?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Owen O'Malley updated HIVE-6060:
--------------------------------
Attachment: h-5317.patch
Rowids are not unique across buckets and thus the unique identifier is thus:
(transaction id, bucket id, row id). Alan suggested offline that I add bucket
id to API so that we aren't forced to maintain the current restriction of one
HDFS file per a bucket. I've also added my thoughts on what the reader would
look like.
I also need to look at what the API looks like for vectorization.
> Define API for RecordUpdater and UpdateReader
> ---------------------------------------------
>
> Key: HIVE-6060
> URL: https://issues.apache.org/jira/browse/HIVE-6060
> Project: Hive
> Issue Type: Sub-task
> Reporter: Owen O'Malley
> Assignee: Owen O'Malley
> Attachments: h-5317.patch, h-5317.patch
>
>
> We need to define some new APIs for how Hive interacts with the file formats
> since it needs to be much richer than the current RecordReader and
> RecordWriter.
--
This message was sent by Atlassian JIRA
(v6.1.4#6159)
