[
https://issues.apache.org/jira/browse/HIVE-12763?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15086084#comment-15086084
]
Alan Gates commented on HIVE-12763:
-----------------------------------
General comment, in patches like this that are huge and 99% generated code it's
helpful to post a version for review that's just the non-generated code.
I didn't review the bit vector code, but here's some feedback for the rest:
In the thrift interface changes, you should add the new bitVector fields as
'optional' rather than required for backwards compatibility.
Does it make sense to make this configurable? When would you not want to use
this? I understand it doesn't work with RDBMS and only with HBase metastore,
but there's already a config for the HBase metastore, so you could just check
that that's set.
We are now using Thrift 0.9.3 to generate the thrift code, not 0.9.0. You'll
need to install 0.9.3 and generate the code.
In hbase_metastore_proto.proto, I'm surprised to see that you are storing the
bit vectors as strings. Why not as bytes?
Since you're adding javolution to the code you'll need to add it to the NOTICE
file.
> Use bit vector to track per partition NDV
> -----------------------------------------
>
> Key: HIVE-12763
> URL: https://issues.apache.org/jira/browse/HIVE-12763
> Project: Hive
> Issue Type: Improvement
> Reporter: Pengcheng Xiong
> Assignee: Pengcheng Xiong
> Attachments: HIVE-12763.01.patch
>
>
> This will improve merging of per partitions stats.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
