[
https://issues.apache.org/jira/browse/HADOOP-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Haohui Mai updated HADOOP-9669:
-------------------------------
Attachment: HADOOP-9669.patch
Here is a more version that utilizes Java's ByteBuffer. It should be more
efficient.
The APIs are compatible with the previous version.
The APIs might not be ideal towards an efficient implementation of XDR
serialization / deserialization. I'm tracking my proposals to the APIs in a
separate JIRA.
> There are multiple array creations and array copies for a single nfs rpc reply
> ------------------------------------------------------------------------------
>
> Key: HADOOP-9669
> URL: https://issues.apache.org/jira/browse/HADOOP-9669
> Project: Hadoop Common
> Issue Type: Improvement
> Components: nfs
> Reporter: Tsz Wo (Nicholas), SZE
> Assignee: Haohui Mai
> Attachments: HADOOP-9669.patch
>
>
> XDR.writeXxx(..) methods ultimately use the static XDR.append(..) for writing
> each data type. The static append creates a new array and copy data.
> Therefore, for a singe reply such as RpcAcceptedReply.voidReply(..), there
> are multiple array creations and array copies. For example, there are at
> least 6 array creations and array copies for RpcAcceptedReply.voidReply(..).
--
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
