MacLochlainns Weblog

Michael McLaughlin's Technical Blog

Site Admin

INSERT Statement

without comments

INSERT Statement

Learning Outcomes

  • Learn how to use positional- and named-notation in INSERT statements.
  • Learn how to use the VALUES clause in INSERT statements.
  • Learn how to use subqueries in INSERT statements.

The INSERT statement lets you enter data into tables and views in two ways: via an INSERT statement with a VALUES clause and via an INSERT statement with a query. The VALUES clause takes a list of literal values (strings, numbers, and dates represented as strings), expression values (return values from functions), or variable values.

Query values are results from SELECT statements that are subqueries (covered earlier in this appendix). INSERT statements work with scalar, single-row, and multiple-row subqueries. The list of columns in the VALUES clause or SELECT clause of a query (a SELECT list) must map to the positional list of columns that defines the table. That list is found in the data dictionary or catalog. Alternatively to the list of columns from the data catalog, you can provide a named list of those columns. The named list overrides the positional (or default) order from the data catalog and must provide at least all mandatory columns in the table definition. Mandatory columns are those that are not null constrained.

Oracle databases differ from other databases in how they implement the INSERT statement. Oracle doesn’t support multiple-row inserts with a VALUES clause. Oracle does support default and override signatures as qualified in the ANSI SQL standards. Oracle also provides a multiple- table INSERT statement. This section covers how you enter data with an INSERT statement that is based on a VALUES clause or a subquery result statement. It also covers multiple-table INSERT statements.

The INSERT statement has one significant limitation: its default signature. The default signature is the list of columns that defines the table in the data catalog. The list is defined by the position and data type of columns. The CREATE statement defines the initial default signature, and the ALTER statement can change the number, data types, or ordering of columns in the default signature.

The default prototype for an INSERT statement allows for an optional column list that overrides the default list of columns. When you provide the column list you choose to implement named-notation, which is the right way to do it. Relying on the insertion order of the columns is a bad idea. An INSERT statement without a list of column names is a position-notation statement. Position-notation is bad because somebody can alter that order and previously written INSERT statements will break or put data in the wrong columns.

Like methods in OOPLs, an INSERT statement without the optional column list constructs an instance (or row) of the table using the default constructor. The override constructor for a row is defined by any INSERT statement when you provide an optional column list. That’s because it overrides the default constructor.

The generic prototype for an INSERT statement is confusing when it tries to capture both the VALUES clause and the result set from a query. Therefore, I’ve opted to provide two generic prototypes.

Insert by value

The first uses the VALUES clause:

INSERT
INTO table_name
[( column1, column2, column3, ...)] VALUES
( value1, value2, value3, ...);

Notice that the prototype for an INSERT statement with the result set from a query doesn’t use the VALUES clause at all. A parsing error occurs when the VALUES clause and query both occur in an INSERT statement.

The second prototype uses a query and excludes the VALUES clause. The subquery may return one to many rows of data. The operative rule is that all columns in the query return the same number of rows of data, because query results should be rectangles—rectangles made up of one to many rows of columns.

Insert by subquery

Here’s the prototype for an INSERT statement that uses a subquery:

INSERT
INTO table_name
[( column1, column2, column3, ...)]
( SELECT value1, value2, value3, ... FROM table_name WHERE ...);

A query, or SELECT statement, returns a SELECT list. The SELECT list is the list of columns, and it’s evaluated by position and data type. The SELECT list must match the definition of the table or the override signature provided.

Default signatures present a risk of data corruption through insertion anomalies, which occur when you enter bad data in tables. Mistakes transposing or misplacing values can occur more frequently with a default signature, because the underlying table structure can change. As a best practice, always use named notation by providing the optional list of values; this should help you avoid putting the right data in the wrong place.

The following subsections provide examples that use the default and override syntax for INSERT statements in Oracle databases. The subsections also cover multiple-table INSERT statements and a RETURNING INTO clause, which is an extension of the ANSI SQL standard. Oracle uses the RETURNING INTO clause to manage large objects, to return autogenerated identity column values, and to support some of the features of Oracle’s dynamic SQL. Note that Oracle also supports a bulk INSERT statement, which requires knowledge of PL/SQL.

Written by maclochlainn

April 5th, 2022 at 1:23 pm