Overview of Commands

In OLE DB, Data Definition Language (DDL) and Data Manipulation Language (DML) statements are referred to as text commands. A command object contains a text command and its main purpose is to execute the command. Commands expose various interfaces representing different areas of functionality of a query processor, including query formulation, preparation, and execution. Text commands are expressed in a provider-specific language—usually ANSI SQL92. Commands can be in one of four states: Initial, Unprepared, Prepared, or Executed. Parameters can be used with commands to bind to consumer variables at execution time. When executed, commands return results: either a rowset or a row count, or multiple results. This chapter discusses the use and states of command objects, binding input parameters and determining support of output parameters, and the results set.

A command is used to execute a provider-specific text command, such as an SQL statement. It is important not to confuse a command, which is an object, and its command text, which is a string. Commands are usually used for data definition, such as creating a table or granting privileges, and data manipulation, such as updating or deleting rows. A special case of data manipulation is creating a rowset—for example, an SQL SELECT statement.

Providers are not required to support commands. In general, providers built on top of a DBMS, such as an SQL DBMS, support commands, and providers built on top of a simple data structure, such as a file or an array of data in an application, do not support commands.

Command objects:

CoType TCommand {
   [mandatory] interface IAccessor;
   [mandatory] interface IColumnsInfo;
   [mandatory] interface ICommand;
   [mandatory] interface ICommandProperties;
   [mandatory] interface ICommandText;
   [mandatory] interface IConvertType;
   [optional]  interface IColumnsRowset;
   [optional]  interface ICommandPrepare;
   [optional]  interface ICommandWithParameters;
   [optional]  interface ISupportErrorInfo;
}