[
https://issues.apache.org/jira/browse/HIVE-1501?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12913723#action_12913723
]
John Sichi commented on HIVE-1501:
----------------------------------
Example:
create table `_t`(`_i` int, `_j` int);
create index x on table `_t`(`_j`) as 'compact' with deferred rebuild;
alter index x on `_t` rebuild;
gives
FAILED: Parse Error: line 1:48 mismatched input ',' expecting CharSetLiteral in
character string literal
To see why, look at
org.apache.hadoop.hive.ql.index.compact.CompactIndexHandler, method
getIndexBuilderMapRedTask. It constructs an internal SQL statement (INSERT)
which populates the index table structure. However, it neglects to quote the
table/column names, leading to invalid syntax. (Hive uses backticks to quote
identifiers with special characters--I think this currently only applies to
leading underscores, but later we'll support arbitrary identifiers.)
HiveUtils.unparseIdentifier should be used for quoting.
> when generating reentrant INSERT for index rebuild, quote identifiers using
> backticks
> -------------------------------------------------------------------------------------
>
> Key: HIVE-1501
> URL: https://issues.apache.org/jira/browse/HIVE-1501
> Project: Hadoop Hive
> Issue Type: Bug
> Components: Indexing
> Affects Versions: 0.7.0
> Reporter: John Sichi
> Assignee: He Yongqiang
> Fix For: 0.7.0
>
>
> Yongqiang, you mentioned that you weren't able to do this due to SORT BY not
> accepting them. The SORT BY is gone now as of HIVE-1494 (and SORT BY needs
> to be fixed anyway).
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
