You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "xiaojin.wy (Jira)" <ji...@apache.org> on 2019/12/18 08:40:00 UTC
[jira] [Updated] (FLINK-15310) A timestamp result get by a select
sql and a csvsink sql is different
[ https://issues.apache.org/jira/browse/FLINK-15310?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
xiaojin.wy updated FLINK-15310:
-------------------------------
Description:
*The sql is:*
CREATE TABLE `orders` (
rowtime TIMESTAMP,
id INT,
product VARCHAR,
units INT
) WITH (
'format.field-delimiter'='|',
'connector.type'='filesystem',
'format.derive-schema'='true',
'connector.path'='/defender_test_data/daily_regression_blink_sql_1.10/test_agg/sources/orders.csv',
'format.type'='csv'
);
select floor(rowtime to hour) as rowtime, count(*) as c from orders group by floor(rowtime to hour)
The result got in a sqlClient environment which use the sql above is like this:
!image-2019-12-18-16-33-24-510.png!
But the same sql write directly to a cvs batch sink will get a result like this:
1972-03-06 08:44:36.736|4
1972-03-06 08:44:36.736|1
was:
*The sql is:*
CREATE TABLE `orders` (
rowtime TIMESTAMP,
id INT,
product VARCHAR,
units INT
) WITH (
'format.field-delimiter'='|',
'connector.type'='filesystem',
'format.derive-schema'='true',
'connector.path'='/defender_test_data/daily_regression_blink_sql_1.10/test_agg/sources/orders.csv',
'format.type'='csv'
);
select floor(rowtime to hour) as rowtime, count(*) as c from orders group by floor(rowtime to hour)
The result got in a sqlClient environment which use the sql above:
!image-2019-12-18-16-33-24-510.png!
But the same sql write directly to a cvs batch sink will get a result like this:
1972-03-06 08:44:36.736|4
1972-03-06 08:44:36.736|1
> A timestamp result get by a select sql and a csvsink sql is different
> ---------------------------------------------------------------------
>
> Key: FLINK-15310
> URL: https://issues.apache.org/jira/browse/FLINK-15310
> Project: Flink
> Issue Type: Bug
> Components: Table SQL / Client
> Affects Versions: 1.10.0
> Environment: execution:
> planner: blink
> type: batch
> Reporter: xiaojin.wy
> Priority: Major
> Attachments: image-2019-12-18-16-33-24-510.png
>
>
> *The sql is:*
> CREATE TABLE `orders` (
> rowtime TIMESTAMP,
> id INT,
> product VARCHAR,
> units INT
> ) WITH (
> 'format.field-delimiter'='|',
> 'connector.type'='filesystem',
> 'format.derive-schema'='true',
> 'connector.path'='/defender_test_data/daily_regression_blink_sql_1.10/test_agg/sources/orders.csv',
> 'format.type'='csv'
> );
> select floor(rowtime to hour) as rowtime, count(*) as c from orders group by floor(rowtime to hour)
> The result got in a sqlClient environment which use the sql above is like this:
> !image-2019-12-18-16-33-24-510.png!
> But the same sql write directly to a cvs batch sink will get a result like this:
> 1972-03-06 08:44:36.736|4
> 1972-03-06 08:44:36.736|1
--
This message was sent by Atlassian Jira
(v8.3.4#803005)