[
https://issues.apache.org/jira/browse/HIVE-26046?focusedWorklogId=800752&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-800752
]
ASF GitHub Bot logged work on HIVE-26046:
-----------------------------------------
Author: ASF GitHub Bot
Created on: 15/Aug/22 22:24
Start Date: 15/Aug/22 22:24
Worklog Time Spent: 10m
Work Description: nrg4878 commented on code in PR #3276:
URL: https://github.com/apache/hive/pull/3276#discussion_r946175305
##########
standalone-metastore/metastore-server/src/main/java/org/apache/hadoop/hive/metastore/dataconnector/jdbc/MySQLConnectorProvider.java:
##########
@@ -90,10 +90,20 @@ protected String getDataType(String dbDataType, int size) {
// map any db specific types here.
switch (dbDataType.toLowerCase())
{
+ case "bit":
+ return toHiveBitType(size);
default:
mappedType = ColumnType.VOID_TYPE_NAME;
break;
}
return mappedType;
}
+
+ private String toHiveBitType(int size) {
+ if (size <= 1) {
+ return ColumnType.BOOLEAN_TYPE_NAME;
+ } else {
+ return ColumnType.BIGINT_TYPE_NAME;
Review Comment:
@zhangbutao Sorry for a delayed response.
So with option #1, HMS would replace the bit type columns for MySQL with
bin(column_name). I kinda like this option in that it simplifies the logic and
pushes the onus on the JDBC datasource natively.
Option #2 uses hive UDF which maybe different from the native
implementation. And it also involves computation on the hive side.
What do you think?
Issue Time Tracking
-------------------
Worklog Id: (was: 800752)
Time Spent: 2.5h (was: 2h 20m)
> MySQL's bit datatype is default to void datatype in hive
> --------------------------------------------------------
>
> Key: HIVE-26046
> URL: https://issues.apache.org/jira/browse/HIVE-26046
> Project: Hive
> Issue Type: Sub-task
> Components: Standalone Metastore
> Affects Versions: 4.0.0
> Reporter: Naveen Gangam
> Assignee: zhangbutao
> Priority: Major
> Labels: pull-request-available
> Time Spent: 2.5h
> Remaining Estimate: 0h
>
> describe on a table that contains a "bit" datatype gets mapped to void. We
> need a explicit conversion logic in the MySQL ConnectorProvider to map it to
> a suitable datatype in hive.
> {noformat}
> +-------------------------------+---------------------------------------------------+----------------------------------------------------+
> | col_name | data_type
> | comment |
> +-------------------------------+---------------------------------------------------+----------------------------------------------------+
> | tbl_id | bigint
> | from deserializer |
> | create_time | int
> | from deserializer |
> | db_id | bigint
> | from deserializer |
> | last_access_time | int
> | from deserializer |
> | owner | varchar(767)
> | from deserializer |
> | owner_type | varchar(10)
> | from deserializer |
> | retention | int
> | from deserializer |
> | sd_id | bigint
> | from deserializer |
> | tbl_name | varchar(256)
> | from deserializer |
> | tbl_type | varchar(128)
> | from deserializer |
> | view_expanded_text | string
> | from deserializer |
> | view_original_text | string
> | from deserializer |
> | is_rewrite_enabled | void
> | from deserializer |
> | write_id | bigint
> | from deserializer
> {noformat}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
