[
https://issues.apache.org/jira/browse/HADOOP-6630?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12844639#action_12844639
]
Allen Wittenauer commented on HADOOP-6630:
------------------------------------------
Actually, I misspoke a bit. It will be full path if it comes from $PATH. If
you call with a relative path, it will contain the relative path.
So to be more explicit:
if I put this code:
{code}
#!/bin/bash
echo $BASH_SOURCE
{code}
in $HOME/bin and make sure $HOME/bin is in my path, you'll see the following:
awittena-mn:bin awittena$ testexec
/Users/awittena/bin/testexec
awittena-mn:~ awittena$ bin/testexec
bin/testexec
awittena-mn:~ awittena$ cd bin && testexec
/Users/awittena/bin/testexec
awittena-mn:~ awittena$ cd bin && ./testexec
./testexec
So the cd+pwd combo is still needed, but this should fix the "in the $PATH"
issue. It is a shame that bash hid the "can't find hadoop-config.sh" error for
the past few years. If you run it with -x set, you'll see it tho.
> hadoop-config.sh fails to get executed if hadoop wrapper scripts are in path
> ----------------------------------------------------------------------------
>
> Key: HADOOP-6630
> URL: https://issues.apache.org/jira/browse/HADOOP-6630
> Project: Hadoop Common
> Issue Type: Bug
> Affects Versions: 0.20.2
> Reporter: Allen Wittenauer
> Attachments: HADOOP-6630.txt
>
>
> If the hadoop/bin commands are in the path, hadoop-config.sh doesn't executed
> until much later in the stack because $0 lacks a path.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
