[
https://issues.apache.org/jira/browse/HADOOP-7236?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13028794#comment-13028794
]
Matt Foley commented on HADOOP-7236:
------------------------------------
It looks generally good, and this refactoring is very good for the FsShell.
However, I think John George was on the right track when he questioned adding
mkdirs(), etc., methods to the PathData class. We seem to already be creating
FileContext as a mirror of all FS methods, we shouldn't do the same with
PathData. To me it seems obvious that the invocation of mkdirs(), rightly
removed from FSCommand, should be located in the Mkdir class.
Daryl and I talked about various ways to try to "simplify the switch to
FileContext". We concluded (or at least I did -- sorry if I'm putting words in
your mouth :-) ) that we were going through contortions to avoid changing
twenty single lines in twenty different files at some point in the future;
i.e., converting the invocations of FileSystem methods to FileContext methods
in each of the implementations of the various FSCommands. While it would be
nice to avoid that future editing, it isn't worth complicating the design, in
my opinion. And isn't it kind of the point of all this that each command (or
related subset of commands) can become a separate subclass of FSCommand?
Thanks.
> Refactor FsShell's mkdir
> ------------------------
>
> Key: HADOOP-7236
> URL: https://issues.apache.org/jira/browse/HADOOP-7236
> Project: Hadoop Common
> Issue Type: Improvement
> Components: fs
> Affects Versions: 0.23.0
> Reporter: Daryn Sharp
> Assignee: Daryn Sharp
> Attachments: HADOOP-7236-2.patch, HADOOP-7236.patch
>
>
> Need to refactor tail to conform to new FsCommand class.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
