[
https://issues.apache.org/jira/browse/HADOOP-12196?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14623433#comment-14623433
]
Kengo Seki commented on HADOOP-12196:
-------------------------------------
bq. I wonder if we are overthinking this and just need to throw stderr away.
It may be enough. If we choose this, I'd like to display some user-friendly
messages that indicates which files are skipped.
bq. Or maybe take the head of the file and see if it:
Sounds good. I'm thinking of leveraging file command for file type detection as
above, and making shellcheck to process it if its type is supported by
shellcheck (that is sh, bash, ksh),
> shellcheck plugin is picking up target executables
> --------------------------------------------------
>
> Key: HADOOP-12196
> URL: https://issues.apache.org/jira/browse/HADOOP-12196
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: yetus
> Affects Versions: HADOOP-12111
> Reporter: Allen Wittenauer
> Assignee: Kengo Seki
> Attachments: HADOOP-12196.HADOOP-12111.01.patch
>
>
> When hadoop-yarn-server-nodemanager is in the build, it generates
> target/native/target/usr/local/bin/(test-)container-executor. shellcheck
> picks these up and throws a "hGetContents: invalid argument (invalid byte
> sequence)" error.
> We need some sort of safety to determine if a file is binary or shell code
> and/or prevent target directories from getting picked up. I suspect the
> former to be more reliable than the latter since we can't assume maven.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
