[
https://issues.apache.org/jira/browse/FLINK-7114?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16077427#comment-16077427
]
mingleizhang edited comment on FLINK-7114 at 7/7/17 12:54 AM:
--------------------------------------------------------------
This is not only the problem of hadoop version, but this kinda hardcoded is not
really a very good way. And I will refine it to
{{DistributedFileSystem.class.getName}} for getting its classpath to a more
flexable way. Similarly, I believe change all this kind code in this JIRA is
not good I believe. So, I would create subtask for every module if there is no
problem.
was (Author: mingleizhang):
This is not only the problem of hadoop version, but this kinda hardcoded way is
not really a very good way. And I will refine it to
{{DistributedFileSystem.class.getName}} for getting its classpath to a more
flexable way. Similarly, I believe change all this kind code in this JIRA is
not good I believe. So, I would create subtask for every module if there is no
problem.
> Remove the hardcoded way like classpath is written by a string
> --------------------------------------------------------------
>
> Key: FLINK-7114
> URL: https://issues.apache.org/jira/browse/FLINK-7114
> Project: Flink
> Issue Type: Bug
> Components: Local Runtime
> Reporter: mingleizhang
> Assignee: mingleizhang
>
> For example, Like in {{HadoopFileSystem}} class, there is a field which is
> written like {{private static final String DEFAULT_HDFS_CLASS =
> "org.apache.hadoop.hdfs.DistributedFileSystem";}}. This way of writing has a
> bad stuff, once the classpath changes, it is difficult to find the error. I
> suggest we should change it to a more flexable way.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
