Github user fhueske commented on a diff in the pull request: https://github.com/apache/flink/pull/5564#discussion_r170271213 --- Diff: flink-libraries/flink-table/src/main/scala/org/apache/flink/table/descriptors/Schema.scala --- @@ -32,11 +32,36 @@ import scala.collection.mutable */ class Schema extends Descriptor { + private var deriveFields: Option[String] = None + // maps a field name to a list of properties that describe type, origin, and the time attribute private val tableSchema = mutable.LinkedHashMap[String, mutable.LinkedHashMap[String, String]]() private var lastField: Option[String] = None + /** + * Derives field names and types from a preceding connector or format. Additional fields that + * are defined in this schema extend the derived fields. The derived fields are + * added in an alphabetical order according to their field name. + */ + def deriveFieldsAlphabetically(): Schema = { --- End diff -- I think we should support inferring the format from the schema rather than the schema from the format. This would be more aligned with how it would work in a `CREATE TABLE` statement and how Hive is doing it for example. We should still support to define the format explicitly though.
---