Skip to content

AlterTableAddColumnsCommand Logical Runnable Command

AlterTableAddColumnsCommand is a LeafRunnableCommand that represents AddColumns logical operator.

Creating Instance

AlterTableAddColumnsCommand takes the following to be created:

  • Table (TableIdentifier)
  • Columns to Add (StructFields)

AlterTableAddColumnsCommand is created when:

Executing Command

run(
  sparkSession: SparkSession): Seq[Row]

run verifyAlterTableAddColumn (with the SessionCatalog).

run uncaches the table.

run requests the SessionCatalog to refreshTable.

run checks the column names (against any duplications) and types, and re-constructs the original schema of columns from their column metadata (if there is any).

run requests the SessionCatalog to alterTableDataSchema.

run is part of the RunnableCommand abstraction.

verifyAlterTableAddColumn

verifyAlterTableAddColumn(
  conf: SQLConf,
  catalog: SessionCatalog,
  table: TableIdentifier): CatalogTable

verifyAlterTableAddColumn asserts that the table is as follows:

  1. The table is not a view
  2. The table is a Hive table or one of the supported FileFormats

verifyAlterTableAddColumn requests the given SessionCatalog for the getTempViewOrPermanentTableMetadata.

verifyAlterTableAddColumn throws an AnalysisException if the table is a VIEW:

ALTER ADD COLUMNS does not support views.
You must drop and re-create the views for adding the new columns. Views: [table]

For a Spark table (that is non-Hive), verifyAlterTableAddColumn finds the implementation of the table provider and makes sure that the table provider is one of the following supported file formats:

Otherwise, verifyAlterTableAddColumn throws an AnalysisException:

ALTER ADD COLUMNS does not support datasource table with type [tableType].
You must drop and re-create the table for adding the new columns. Tables: [table]