[
https://issues.apache.org/jira/browse/FLINK-35392?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Glauber M Dantas updated FLINK-35392:
-------------------------------------
Description:
When transforming a DataStream to a Table and then using that table in a Join
Lookup, an exception is raised: {{{}"Temporal table join currently only
supports 'FOR SYSTEM_TIME AS OF' left table's time attribute field"{}}}.
>From what I understood from the issue, it happens as the planner expects that
>the Tables part of the Join are registered previously on the Catalog.
The solution was to avoid using the DataStream to Table conversion:
```
Table kinesisStreamTbl = tblEnv.fromDataStream(
kinesisDataStream,
Schema.newBuilder()
// "proc_time" would be needed to make the lookup using
'FOR SYSTEM_TIME AS OF'; see:
[https://nightlies.apache.org/flink/flink-docs-master/docs/dev/table/sql/queries/joins/#processing-time-temporal-join]
.columnByExpression("proc_time", "PROCTIME()")
.build()
);
```
And instead create and use a Table registered on the Catalog:
```
tblEnv.createTable("KinesisInput", TableDescriptor.forConnector("kinesis")...);
```
If you need more details, you may access the discussion at:
[https://stackoverflow.com/questions/78485099/sedona-flink-sql-lookup-on-external-database-failing-when-using-for-system-time]
was:
When transforming a DataStream to a Table and then using that table in a Join
Lookup, an exception is raised: {{{}"Temporal table join currently only
supports 'FOR SYSTEM_TIME AS OF' left table's time attribute field"{}}}.
>From what I understood from the issue, it happens as the planner expects that
>the Tables part of the Join are registered previously on the Catalog.
The solution was to avoid using the DataStream to Table conversion:
```
Table kinesisStreamTbl = tblEnv.fromDataStream(
kinesisDataStream,
Schema.newBuilder()
// "proc_time" would be needed to make the lookup using
'FOR SYSTEM_TIME AS OF'; see:
https://nightlies.apache.org/flink/flink-docs-master/docs/dev/table/sql/queries/joins/#processing-time-temporal-join
.columnByExpression("proc_time", "PROCTIME()")
.build()
);
```
And instead create and use a Table registered on the Catalog:
```
tblEnv.createTable("KinesisInput", TableDescriptor.forConnector("kinesis")...);
```
If you need more details, you may access the discussion at:
https://stackoverflow.com/questions/78485099/sedona-flink-sql-lookup-on-external-database-failing-when-using-for-system-time
> SQL Lookup Join on external database failing when using 'FOR SYSTEM_TIME AS
> OF' statement
> -----------------------------------------------------------------------------------------
>
> Key: FLINK-35392
> URL: https://issues.apache.org/jira/browse/FLINK-35392
> Project: Flink
> Issue Type: Bug
> Components: Table SQL / Planner
> Affects Versions: 1.18.1
> Reporter: Glauber M Dantas
> Priority: Major
>
> When transforming a DataStream to a Table and then using that table in a Join
> Lookup, an exception is raised: {{{}"Temporal table join currently only
> supports 'FOR SYSTEM_TIME AS OF' left table's time attribute field"{}}}.
> From what I understood from the issue, it happens as the planner expects that
> the Tables part of the Join are registered previously on the Catalog.
> The solution was to avoid using the DataStream to Table conversion:
>
> ```
> Table kinesisStreamTbl = tblEnv.fromDataStream(
> kinesisDataStream,
> Schema.newBuilder()
> // "proc_time" would be needed to make the lookup
> using 'FOR SYSTEM_TIME AS OF'; see:
> [https://nightlies.apache.org/flink/flink-docs-master/docs/dev/table/sql/queries/joins/#processing-time-temporal-join]
> .columnByExpression("proc_time", "PROCTIME()")
> .build()
> );
> ```
> And instead create and use a Table registered on the Catalog:
>
> ```
> tblEnv.createTable("KinesisInput",
> TableDescriptor.forConnector("kinesis")...);
> ```
>
> If you need more details, you may access the discussion at:
> [https://stackoverflow.com/questions/78485099/sedona-flink-sql-lookup-on-external-database-failing-when-using-for-system-time]
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
