[ https://issues.apache.org/jira/browse/FLINK-3754?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15253874#comment-15253874 ]
ASF GitHub Bot commented on FLINK-3754: --------------------------------------- Github user yjshen commented on the pull request: https://github.com/apache/flink/pull/1916#issuecomment-213419043 The type annotation work is done from bottom to top: Firstly, we know each schema of the two input, and we know `List[] expression` in `Project` are used to manipulate one row of table data as input and output one value per expression, therefore, we can infer the the output schema of `Project` (in the current impl this was expressed as: `def output: Seq[Attribute]`) if we know each expressions `dataType`. For example, `Add`'s dataType is same as it's input, `Or`'s dataType is always `Boolean`, `pow(a, b)`'s dataType is always `Double`, however, if and only if we understand all kinds of expressions, we are able to infer its `dataType`. The main problems here is we only have `Call`(Unresolved Function) generated during expression construction, therefore, we should resolve them first into solid `Expression`s. `FunctionCatalog` is introduced here for a mapping from `FunctionName -> Expression`, we can easily finish the translation work as we look up `catalog`. > Add a validation phase before construct RelNode using TableAPI > -------------------------------------------------------------- > > Key: FLINK-3754 > URL: https://issues.apache.org/jira/browse/FLINK-3754 > Project: Flink > Issue Type: Improvement > Components: Table API > Affects Versions: 1.0.0 > Reporter: Yijie Shen > Assignee: Yijie Shen > > Unlike sql string's execution, which have a separate validation phase before > RelNode construction, Table API lacks the counterparts and the validation is > scattered in many places. > I suggest to add a single validation phase and detect problems as early as > possible. -- This message was sent by Atlassian JIRA (v6.3.4#6332)