[
https://issues.apache.org/jira/browse/HIVE-28792?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17931113#comment-17931113
]
Krisztian Kasa commented on HIVE-28792:
---------------------------------------
The issue is caused by that the literal \{{'A'}} in the case expression is
converted to char(3) and padded:
{code}
CBO PLAN:
HiveProject(col1=[$0], col2=[CAST(CASE(=(UPPER($0), _UTF-16LE'A '),
_UTF-16LE'OK':VARCHAR(2147483647) CHARACTER SET "UTF-16LE",
_UTF-16LE'N/A':VARCHAR(2147483647) CHARACTER SET
"UTF-16LE")):VARCHAR(2147483647) CHARACTER SET "UTF-16LE"])
HiveTableScan(table=[[default, testchar]], table:alias=[testchar])
{code}
{code}
STAGE DEPENDENCIES:
Stage-0 is a root stage
STAGE PLANS:
Stage: Stage-0
Fetch Operator
limit: -1
Processor Tree:
TableScan
alias: testchar
Select Operator
expressions: col1 (type: char(3)), if((upper(col1) = 'A '), 'OK',
'N/A') (type: string)
outputColumnNames: _col0, _col1
ListSink
{code}
However at the Hive operator level in the expression
{code}
upper(col1) = 'A '
{code}
The function call {{upper(col1)}} has {{char(3)}} type and the literal {{'A
'}} has {{string}} type. The result of the function is converted to string and
during the conversion trailing spaces are trimmed.
> Wrong results when query has function call with char parameter type in case
> expression
> --------------------------------------------------------------------------------------
>
> Key: HIVE-28792
> URL: https://issues.apache.org/jira/browse/HIVE-28792
> Project: Hive
> Issue Type: Bug
> Components: CBO
> Reporter: Krisztian Kasa
> Priority: Major
>
> Repro
> {code:java}
> CREATE EXTERNAL TABLE testchar (col1 char(3));
> INSERT INTO testchar VALUES ('A'),('B');
> select col1, case upper(col1) when 'A' then 'OK' else 'N/A' end as col2 from
> testchar;
> {code}
> The query should return
> {code:java}
> A OK
> B N/A
> {code}
> but it returns
> {code:java}
> A N/A
> B N/A
> {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
