[
https://issues.apache.org/jira/browse/HIVE-12067?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14949161#comment-14949161
]
Gopal V commented on HIVE-12067:
--------------------------------
This doesn't seem to be a bug in ORC, it is a bug in LazySimpleSerDe
encoding/decoding - it should ideally be writing \\N and reading out the same
data as ORC.
"\N" is a valid string & is not null.
Try
{{select count(1) from (select '\N' as a from table) tbl where tbl.a is null;}}
> ORC read nullString(eg: \N) columns, can't return NULL
> ------------------------------------------------------
>
> Key: HIVE-12067
> URL: https://issues.apache.org/jira/browse/HIVE-12067
> Project: Hive
> Issue Type: Bug
> Affects Versions: 1.2.1
> Reporter: Bill Wailliam
> Assignee: Ashutosh Chauhan
>
> Text format:
> SQL:create table test_input as select '\\N' from table;
> hive> select * from test_input limit 3;
> OK
> NULL
> NULL
> NULL
> ===============
> ORC format:
> set hive.default.fileformat=Orc;
> create table test_orc_input as select '\\N' from table;
> hive> select * from test_orc_input limit 3;
> OK
> \N
> \N
> \N
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
