hadoop-hive-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ashish Thusoo (JIRA)" <j...@apache.org>
Subject [jira] Commented: (HIVE-226) SerDe should support "null" column values
Date Wed, 11 Mar 2009 00:22:50 GMT

    [ https://issues.apache.org/jira/browse/HIVE-226?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12680701#action_12680701
] 

Ashish Thusoo commented on HIVE-226:
------------------------------------

This works with \N in the load file for nulls. We have a test for that - input21.q.

> SerDe should support "null" column values
> -----------------------------------------
>
>                 Key: HIVE-226
>                 URL: https://issues.apache.org/jira/browse/HIVE-226
>             Project: Hadoop Hive
>          Issue Type: Improvement
>          Components: Serializers/Deserializers
>            Reporter: Josh Ferguson
>
> Currently if you attempt to load data that has null values it will work but selecting
data back out of the table will fail due to a null pointer exception during deserialization.
> Suppose we have a generic users table with ^A separated fields.
> CREATE TABLE users 
> (id STRING, properties MAP<STRING, STRING>) 
> ROW FORMAT DELIMITED 
> COLLECTION ITEMS TERMINATED BY '44'
> MAP KEYS TERMINATED BY '58'
> STORED AS TEXTFILE;
> we might insert this data (where spaces are ^A characters)
> 1 key:value
> 2 
> 3 key:value
> Then the following queries will fail
> SELECT id FROM users;
> SELECT id, properties FROM users;
> SELECT properties FROM users;
> But they should not fail.
> Anytime the field delimiter is encountered twice in a row or the field delimiter is encountered
followed immediately by the line delimiter a non-existent value should be assumed for the
appropriate column.
> To circumvent this in my application I have been substituting my own "reserved" world
NULL and the key/value pair NULL:NULL to indicate to my application that particular fields
currently have no value.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


Mime
View raw message