[
https://issues.apache.org/jira/browse/FLINK-6841?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16057610#comment-16057610
]
Fabian Hueske commented on FLINK-6841:
--------------------------------------
To be honest, I don't see a significant improvement by removing
{{StreamTableSourceTable}}.
We have to move the code of {{getRowType()}} from {{StreamTableSourceTable}}
and additionally add a condition that it is only called if the table source is
a {{StreamTableSource}}.
Maybe we even have to add a {{StreamTableSourceTable}} back if both table
sources differ more at some point.
Therefore, I would rather keep the current design as it is for now.
In general, we should think about the design of table sources and improve the
design to have a good design that supports watermarks, time attributes, nested
data, streaming & batch, projection & filter push down, etc. Once we have that,
we might also need to refactor the internals a bit.
> using TableSourceTable for both Stream and Batch OR remove useless import
> -------------------------------------------------------------------------
>
> Key: FLINK-6841
> URL: https://issues.apache.org/jira/browse/FLINK-6841
> Project: Flink
> Issue Type: Improvement
> Components: Table API & SQL
> Affects Versions: 1.4.0
> Reporter: sunjincheng
> Assignee: sunjincheng
>
> 1. {{StreamTableSourceTable}} exist useless import of {{TableException}}
> 2. {{StreamTableSourceTable}} only override {{getRowType}} of
> {{FlinkTable}}, I think we can override the method in {{TableSourceTable}},
> If so we can using {{TableSourceTable}} for both {{Stream}} and {{Batch}}.
> What do you think? [~fhueske] [~twalthr]
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
