[
https://issues.apache.org/jira/browse/FLINK-11142?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16727046#comment-16727046
]
Fabian Hueske commented on FLINK-11142:
---------------------------------------
Hi [~karmagyz],
I'm not sure if this really an issue.
The schema of the table is explicitly specified. If a user specifies a schema
of three fields on an input data set or stream with 5 fields, the result will
be a Table with three fields. If the user needs more fields they can be easily
added to the schema. I think this is quite explicit and well defined behavior.
We can of course add a check and throw an exception, but it would 1) break
existing behavior and 2) force users to declare all fields even though these
are not needed.
> Undefined behavior in the conversion from DataStream/DataSet to Table
> ---------------------------------------------------------------------
>
> Key: FLINK-11142
> URL: https://issues.apache.org/jira/browse/FLINK-11142
> Project: Flink
> Issue Type: Bug
> Components: Table API & SQL
> Affects Versions: 1.7.0
> Reporter: Yangze Guo
> Assignee: Hequn Cheng
> Priority: Minor
>
> When we try to convert DataStream/DataSet to Table. There are two ways of
> adding schema information, *ByName* or *ByPosition*.
> This feature first proposed in [this
> pr|https://github.com/apache/flink/pull/5132].
> In *ByPosition* mode, the current code does not check if the number of fields
> less than its in DataStream/DataSet. This may cause undefined behavior, e.g.
> make a projection in *ByPosition* mode.
> We can either fix it by adding some checking or regard this as a feature and
> just improve the doc to clarify it. In my opinion, the latter way seems
> better.
> [~twalthr] Could you take a look at it when you free?
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
