[ 
https://issues.apache.org/jira/browse/HIVE-15582?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15820751#comment-15820751
 ] 

Jesus Camacho Rodriguez commented on HIVE-15582:
------------------------------------------------

[~bslim], I answer to your comments:
- I even did not realized about the changes that you mention, but they are just 
formatting: it happened because those lines did not have the correct 
indentation.
- We do not need any change for _deserialize_, since the type inferred for 
Druid sources is either _long_ or _float_ (not _byte_, _short_, _int_, or 
_double_).
- DruidOutputFormat already supports byte/short/int (L135-L138 in 
DruidOutputFormat.java).
- Tests for deserializer are already present in TestDruidSerDe. I will upload a 
new patch shortly that adds tests for serializer too.

> Druid CTAS should support BYTE/SHORT/INT types
> ----------------------------------------------
>
>                 Key: HIVE-15582
>                 URL: https://issues.apache.org/jira/browse/HIVE-15582
>             Project: Hive
>          Issue Type: Sub-task
>          Components: Druid integration
>    Affects Versions: 2.2.0
>            Reporter: Jesus Camacho Rodriguez
>            Assignee: Jesus Camacho Rodriguez
>         Attachments: HIVE-15582.patch
>
>
> Currently these types are not recognized and we throw an exception when we 
> try to create a table with them.
> {noformat}
> Caused by: org.apache.hadoop.hive.serde2.SerDeException: Unknown type: INT
>       at 
> org.apache.hadoop.hive.druid.serde.DruidSerDe.serialize(DruidSerDe.java:414)
>       at 
> org.apache.hadoop.hive.ql.exec.FileSinkOperator.process(FileSinkOperator.java:715)
>       ... 22 more
> {noformat}



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to