[
https://issues.apache.org/jira/browse/HDFS-5125?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13747772#comment-13747772
]
Chris Nauroth commented on HDFS-5125:
-------------------------------------
bq. Does this mean we need more than one curWorkingDirectory to support the
right behavior on Windows?
No, there is still one and only one current working directory for the running
process, so there is no ambiguity about a relative path like "dir1\dir2\file3".
The per-drive location is an additional thing that only comes up when you
start using paths that contain a drive spec, like "C:" or "D:", or even
"D:foo.txt" which means "foo.txt in whatever is the current folder of D". This
definitely took me some getting used to when transitioning from Linux. :-)
bq. I refactored the code in part of HADOOP-9897.
Thanks! That makes sense. I'll take a look.
> TestCreateEditsLog#testCanLoadCreatedEditsLog fails on Windows in globStatus
> ----------------------------------------------------------------------------
>
> Key: HDFS-5125
> URL: https://issues.apache.org/jira/browse/HDFS-5125
> Project: Hadoop HDFS
> Issue Type: Bug
> Components: test
> Affects Versions: 3.0.0, 2.1.1-beta
> Reporter: Chris Nauroth
> Assignee: Binglin Chang
>
> This test calls the {{CreateEditsLog}} tool, then runs {{globStatus}} on the
> local file system to find the resulting files before moving them to a
> directory where a NameNode can start up and read them. The HADOOP-9877 patch
> has caused this test to start failing on Windows due to internal calls to
> {{FileContext#getFileLinkStatus}} rejecting the {{Path}} arguments.
--
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
