[
https://issues.apache.org/jira/browse/HIVE-11925?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14904861#comment-14904861
]
Ashutosh Chauhan commented on HIVE-11925:
-----------------------------------------
I think removing format check is alright to do since you likely have made it
sure that your table definition is consistent with data. Anyhow, without this
patch, there is no way to load from named pipe, so it atleast enables loading
data, albeit without a check.
+1
> Hive file format checking breaks load from named pipes
> ------------------------------------------------------
>
> Key: HIVE-11925
> URL: https://issues.apache.org/jira/browse/HIVE-11925
> Project: Hive
> Issue Type: Bug
> Reporter: Sergey Shelukhin
> Assignee: Sergey Shelukhin
> Attachments: HIVE-11925.patch
>
>
> Opening the file and mucking with it when hive.fileformat.check is true (the
> default) breaks the LOAD command from a named pipe. Right now, it's done for
> all the text files blindly to see if they might be in some other format.
> Files.getAttribute can be used to figure out if the input is a named pipe (or
> a socket) and skip the format check.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
