[
https://issues.apache.org/jira/browse/HIVE-12381?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15029434#comment-15029434
]
Pengcheng Xiong commented on HIVE-12381:
----------------------------------------
I have removed hashing and also a configuration for the max length of prefix.
But for encoding, I think we still need that because we need to escape the "/"
otherwise the location in HDFS will be wrong. The path separator problem is
mainly due to the Utilities.join(String... elements). The elements may or may
not contain path separator. It is a consistency problem and it is not that
simple as I investigated. I think it is worth another JIRA to fix it. What do
you think [~ashutoshc]?
> analyze table compute stats for table with special characters will wipe out
> all the table stats
> -----------------------------------------------------------------------------------------------
>
> Key: HIVE-12381
> URL: https://issues.apache.org/jira/browse/HIVE-12381
> Project: Hive
> Issue Type: Bug
> Environment:
> Reporter: Pengcheng Xiong
> Assignee: Pengcheng Xiong
> Attachments: HIVE-12381.01.patch, HIVE-12381.02.patch,
> HIVE-12381.03.patch, HIVE-12381.04.patch
>
>
> repo:
> {code}
> drop table `t//`;
> create table `t//` (col string);
> insert into `t//` values(1);
> insert into `t//` values(null);
> analyze table `t//` compute statistics;
> explain select * from `t//`;
> {code}
> The result
> {code}
> Statistics: Num rows: 1 Data size: 5 Basic stats: COMPLETE Column stats: NONE
> {code}
> is wrong
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
