[
https://issues.apache.org/jira/browse/HADOOP-6949?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12908036#action_12908036
]
Todd Lipcon commented on HADOOP-6949:
-------------------------------------
Hi Navis,
Please format your change as a "patch" file that can be applied against the
trunk of the hadoop-common repository. This one has a diff of 1.txt and 2.txt
which makes it more difficult to apply.
Regarding the change, it seems good except that it's backwards-incompatible for
ObjectWritable. Since ObjectWritable is a public facing class, it's likely that
some people have stored them on disk (eg in sequencefiles) so I don't think we
can make a breaking change like this.
Do you have benchmarks that show a significant improvement by making this
change? If so, it may be worth creating a new CompactObjectWritable that's used
only in the RPC layer, for example, and marked as a Private Evolving interface.
If the improvement is not significant, we should skip this change since we'll
eventually move to a more compact RPC format (Avro) anyway.
> Reduces RPC packet size for primitive arrays, especially long[], which is
> used at block reporting
> -------------------------------------------------------------------------------------------------
>
> Key: HADOOP-6949
> URL: https://issues.apache.org/jira/browse/HADOOP-6949
> Project: Hadoop Common
> Issue Type: Improvement
> Components: io
> Reporter: Navis
> Priority: Trivial
> Attachments: ObjectWritable.diff
>
> Original Estimate: 0.17h
> Remaining Estimate: 0.17h
>
> Current implementation of oah.io.ObjectWritable marshals primitive array
> types as general object array ; array type string + array length + (element
> type string + value)*n
> It would not be needed to specify each element types for primitive arrays.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
