[
https://issues.apache.org/jira/browse/HADOOP-5596?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12698638#action_12698638
]
Tsz Wo (Nicholas), SZE commented on HADOOP-5596:
------------------------------------------------
> It would be more comfortable for users to use EnumSet directly.
Why it would be more comfortable? In the FileSystem, there are many classes
like FileStatus, BlockLocation, etc., implementing the Writable interface.
> Make ObjectWritable support EnumSet
> -----------------------------------
>
> Key: HADOOP-5596
> URL: https://issues.apache.org/jira/browse/HADOOP-5596
> Project: Hadoop Core
> Issue Type: Improvement
> Components: io
> Reporter: He Yongqiang
> Assignee: He Yongqiang
> Attachments: Hadoop-5596-2009-03-31.patch,
> Hadoop-5596-2009-04-03.patch
>
>
> This is a demand for Hadoop-5438.
> Also another small improvement is that i saw that in the beginning of
> readObject, it tries to get the class from PRIMITIVE_NAMES and then conf.
> Maybe it is better to add a direct load after them if the delaredClass is
> still null. Like this:
> {code}
> String className = UTF8.readString(in);
> Class<?> declaredClass = PRIMITIVE_NAMES.get(className);
> if (declaredClass == null) {
> try {
> declaredClass = conf.getClassByName(className);
> } catch (Exception e) {
> }
> }
>
> if(declaredClass == null) {
> try {
> declaredClass = Class.forName(className);
> } catch (ClassNotFoundException e) {
> throw new RuntimeException("readObject can't find class " +
> className, e);
> }
> }
> {code}
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
