[
https://issues.apache.org/jira/browse/HADOOP-11875?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15127974#comment-15127974
]
Haohui Mai commented on HADOOP-11875:
-------------------------------------
bq. the loss of features in the UI (they are sticking to 2.4.1's "legacy mode")
Which feature is lost? Please feel free to file a jira to track this.
bq. "tossed over the fence" WebHDFS authentication that we had to rework to
even make security function if one was using a custom plug-in.
This is an unfair statement. The JSP UI has a series of security issues (e.g.,
XSS) because of its architecture. The new UI is much more secure as these
issues are addressed systematically. Please familiar yourself with relevant
jiras.
bq. The switch took a lot of effort to even get to stable and it's still not
what I would consider feature complete compared to the old API.
The same argument applies to any components. For example, how long does Yetus
take to stabilize? Do you consider Yetus feature complete compared to the
test-patch.sh? Yetus is a good improvement in terms of usability and
maintainability. It also brings in more people in the community. So does the
HTML5 UI.
Going back to the original topic, are there any specific reasons to not fixing
code that fails to compile under the current JDK? What is a good answer to "why
a stable release of Hadoop can't be compiled out-of-the-box?"
> [JDK8] Renaming _ as a one-character identifier to another identifier
> ---------------------------------------------------------------------
>
> Key: HADOOP-11875
> URL: https://issues.apache.org/jira/browse/HADOOP-11875
> Project: Hadoop Common
> Issue Type: Bug
> Reporter: Tsuyoshi Ozawa
> Labels: newbie
> Attachments: build_error_dump.txt
>
>
> From JDK8, _ as a one-character identifier is disallowed. Currently Web UI
> uses it. We should fix them to compile with JDK8.
> https://bugs.openjdk.java.net/browse/JDK-8061549
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
