[
https://issues.apache.org/jira/browse/TAJO-379?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13839844#comment-13839844
]
Hyunsik Choi commented on TAJO-379:
-----------------------------------
It's a nice suggestion, and actually it is necessary feature. However, this
will cause much performance degrade even though CSV files are already well
formatted. So, I would like to suggest that default parsing policy just pus a
NULL value to a column value when any kind of exception occurs. Since this
policy is the same in Apache Hive, this may be not problematic. Later, this
problem might be solved if Tajo has a pluggable CSV file parser system. Thanks!
> Fix Fail to Parse Integer Type in LazyTuple
> -------------------------------------------
>
> Key: TAJO-379
> URL: https://issues.apache.org/jira/browse/TAJO-379
> Project: Tajo
> Issue Type: Bug
> Reporter: DaeMyung Kang
> Assignee: DaeMyung Kang
> Attachments: TAJO-379.patch
>
>
> Situation:
> in LazyTuple
> it just tokenize string with ',' so Long.valueof or Integer.valueof will fail
> when parameter has whitespace.
> Fix:
> trim all tokenized string.
--
This message was sent by Atlassian JIRA
(v6.1#6144)
