[
https://issues.apache.org/jira/browse/HADOOP-8953?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13487806#comment-13487806
]
Daryn Sharp commented on HADOOP-8953:
-------------------------------------
bq. At this point, we'd prefer to work towards full test pass on Windows with
minimal changes to the current trunk code, and then refactor slowly towards the
ideal state after we have those tests as a safety net. This would be a
conservative approach towards improving the codebase's path handling logic.
I generally agree that minimal change is best. I may not understand the whole
problem here, but it appears the cart is dragging the horse? Ie. Are moving in
the wrong direction with {{FileSystem}} to appease a test? Unfortunately the
original patch was deleted (please don't do that, just post another) so I can't
refer back the problem in its entirety.
Note that the change to the {{PathData}} test is breaking what it's trying to
test: path in = path out. I spent a lot of effort trying to ensure that the
output path matches the input to allow scripts to easily parse the output when
multiple args are given.
> Shell PathData parsing failures on Windows
> ------------------------------------------
>
> Key: HADOOP-8953
> URL: https://issues.apache.org/jira/browse/HADOOP-8953
> Project: Hadoop Common
> Issue Type: Bug
> Components: fs
> Affects Versions: trunk-win
> Reporter: Chris Nauroth
> Assignee: Arpit Agarwal
> Attachments: HADOOP-8953-branch-trunk-win.patch
>
>
> Several test suites fail on Windows, apparently due to Windows-specific path
> parsing bugs in PathData.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
