[
https://issues.apache.org/jira/browse/HIVE-6048?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13854634#comment-13854634
]
Ashutosh Chauhan commented on HIVE-6048:
----------------------------------------
We need to interact with {{FileSystem}} objects of Hadoop, which uses {{Path}}
so after patch we need conversions from URI to paths, which earlier use to be
conversion from string to Path. This is an improvement, but I think if in all
these places like CopyWork, LoadTableDesc etc. if we use path objects to hold
location instead of uris, we can avoid those conversions and are consistent
throughout the system.
> Hive load data command rejects file with '+' in the name
> --------------------------------------------------------
>
> Key: HIVE-6048
> URL: https://issues.apache.org/jira/browse/HIVE-6048
> Project: Hive
> Issue Type: Bug
> Components: Query Processor
> Affects Versions: 0.12.0
> Reporter: Xuefu Zhang
> Assignee: Xuefu Zhang
> Attachments: HIVE-6048.1.patch, HIVE-6048.2.patch, HIVE-6048.patch
>
>
> '+' is a valid character in a file name on linux and HDFS. However, loading
> data from such a file into table results the following error:
> {code}
> hive> load data local inpath '/home/xzhang/temp/t+est.txt' into table test;
> Copying data from file:/home/xzhang/temp/t est.txt
> No files matching path: file:/home/xzhang/temp/t est.txt
> FAILED: Execution Error, return code 3 from
> org.apache.hadoop.hive.ql.exec.CopyTask
> {code}
--
This message was sent by Atlassian JIRA
(v6.1.4#6159)
