Yun Zhao created SPARK-10837:
--------------------------------
Summary: TimeStamp could not work on sparksql very well
Key: SPARK-10837
URL: https://issues.apache.org/jira/browse/SPARK-10837
Project: Spark
Issue Type: Bug
Components: SQL
Affects Versions: 1.5.0
Reporter: Yun Zhao
create a file as follows：
{quote}
2015-09-02 09:06:00.000
2015-09-02 09:06:00.001
2015-09-02 09:06:00.100
2015-09-02 09:06:01.000
{quote}
Then upload it to hdfs, for example,put it to /test/testTable.
create table：
{quote}
CREATE EXTERNAL TABLE `testTable`(`createtime` timestamp) LOCATION
'/test/testTable';
{quote}
process sqls：
{quote}
select * from testTable where createtime = "2015-09-02 09:06:00.000";
select * from testTable where createtime > "2015-09-02 09:06:00.000";
select * from testTable where createtime >= "2015-09-02 09:06:00.000";
{quote}
The set of ">=" is not union set of "=" and ">".
but if process sqls as follows：
{quote}
select * from testTable where createtime = timestamp("2015-09-02 09:06:00.000");
select * from testTable where createtime > timestamp("2015-09-02 09:06:00.000");
select * from testTable where createtime >= timestamp("2015-09-02
09:06:00.000");
{quote}
There's no such former problem.
User *explain extended* to find the difference of sqls:
When uses "=","2015-09-02 09:06:00.000" is transfered to timestamp.
When uses ">" or ">=",createtime is transfered to String.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
