[
https://issues.apache.org/jira/browse/HIVE-723?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12739915#action_12739915
]
Zheng Shao commented on HIVE-723:
---------------------------------
A second thought on this: If this feature is in, and the source table schema is
changing, then we might have inserted wrong data into the table.
If the source table schema changed, then most probably users will need to
modify the transform script any way, so it seems to me the overhead of
modifying the query is not that big.
Also it might be a bit hard to do because we still require "AS columns" if it's
an "INSERT OVERWRITE DIRECTORY", so we need to treat these differently.
> output columns should be redundant if an insert immediately follows
> -------------------------------------------------------------------
>
> Key: HIVE-723
> URL: https://issues.apache.org/jira/browse/HIVE-723
> Project: Hadoop Hive
> Issue Type: New Feature
> Components: Query Processor
> Reporter: Namit Jain
>
> If the reducer transformation function is inserting into a hive table
> immediately, hive should be automatically able to infer the number of columns
> instead of asking the user to do so.
> It becomes a problem if the source table schema is changing - the query needs
> to be unnecessarily modified
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
