[
https://issues.apache.org/jira/browse/HIVE-13372?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15215062#comment-15215062
]
Lin Liu commented on HIVE-13372:
--------------------------------
Thanks, Pengcheng.
Just FYI:
After some investigations, we found that the error is not related to sort by,
because even without SORTBY clause, if the data size of the table is
sufficiently large to launch a MR job, the result is still wrong.
> Hive Macro overwritten when multiple macros are used in one column
> ------------------------------------------------------------------
>
> Key: HIVE-13372
> URL: https://issues.apache.org/jira/browse/HIVE-13372
> Project: Hive
> Issue Type: Bug
> Components: Hive
> Affects Versions: 1.2.1
> Reporter: Lin Liu
> Assignee: Pengcheng Xiong
> Priority: Critical
>
> When multiple macros are used in one column, results of the later ones are
> over written by that of the first.
> For example:
> Suppose we have created a table called macro_test with single column x in
> STRING type, and with data as:
> "a"
> "bb"
> "ccc"
> We also create three macros:
> {code}
> CREATE TEMPORARY MACRO STRING_LEN(x string) length(x);
> CREATE TEMPORARY MACRO STRING_LEN_PLUS_ONE(x string) length(x)+1;
> CREATE TEMPORARY MACRO STRING_LEN_PLUS_TWO(x string) length(x)+2;
> {code}
> When we ran the following query,
> {code}
> SELECT
> CONCAT(STRING_LEN(x), ":", STRING_LEN_PLUS_ONE(x), ":",
> STRING_LEN_PLUS_TWO(x)) a
> FROM macro_test
> SORT BY a DESC;
> {code}
> We get result:
> 3:3:3
> 2:2:2
> 1:1:1
> instead of expected:
> 3:4:5
> 2:3:4
> 1:2:3
> Currently we are using Hive 1.2.1, and have applied both HIVE-11432 and
> HIVE-12277 patches.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
