[
https://issues.apache.org/jira/browse/FLINK-32252?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17729559#comment-17729559
]
Jark Wu commented on FLINK-32252:
---------------------------------
[~tanjialiang] That works as expected. Currently, streaming mode assumes the
source is unbounded, so it should never return 0. But I agree the semantics are
not consistent between batch and streaming. We have FLINK-12215 and FLINK-18365
to track and improve this.
> SELECT COUNT(*) will return nothing when source no data return
> --------------------------------------------------------------
>
> Key: FLINK-32252
> URL: https://issues.apache.org/jira/browse/FLINK-32252
> Project: Flink
> Issue Type: Bug
> Components: Table SQL / Planner
> Affects Versions: jdbc-3.1.0
> Reporter: tanjialiang
> Priority: Major
>
>
> mysql source
> {code:java}
> CREATE TABLE student(
> id int primary key auto_increment,
> name varchar(32),
> age int
> );
> INSERT INTO student(name, age) VALUES
> ('tanjl',18),('jark',20),('mike',16),('rose',21);{code}
>
> Flink SQL
> {code:java}
> CREATE TABLE student (
> `id` INT PRIMARY KEY,
> `name` STRING,
> `age` INT
> ) WITH (
> 'connector' = 'jdbc',
> 'url' = 'jdbc:mysql://localhost/test?serverTimezone=UTC',
> 'username' = 'root',
> 'password' = 'root',
> 'table-name' = 'student'
> );
> SELECT count(*) FROM student WHERE age < 15;{code}
> flink will return nothing because jdbc connector push the filter down（after
> flink-connector-jdbc-3.1.0）, which make source no data return.
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
