MDAC 2.5 SDK - ODBC Programmer's Reference
Chapter 17: Programming Considerations
SQL Unicode data types are provided to describe data that resides in Unicode natively on the DBMS. A C Unicode data type is provided to allow an application to bind data to a Unicode buffer. The Driver Manager can convert data from a Unicode C type (SQL_C_WCHAR) to make it function with an ANSI driver.
An ODBC 3.0 or 2.x application will always bind to the ANSI data types. For optimum performance, an ODBC 3.5 (or higher) application should bind to the ANSI data C type if the SQL column type is ANSI, and should bind to the Unicode C data type if the SQL column type is Unicode.
The SQL Unicode type indicators are SQL_WCHAR, SQL_WVARCHAR, and SQL_WLONGVARCHAR. SQL_WCHAR data has a fixed string length, while SQL_WVARCHAR has a variable length with a declared maximum and SQL_WLONGVARCHAR has a variable length with a maximum that depends on the data source.
The C Unicode type indicator is SQL_C_WCHAR. This is the default for each of the SQL Unicode type indicators. All of the SQL types can be converted to SQL_C_WCHAR, and SQL_C_WCHAR can be converted to all of the SQL types. An application can retrieve data in one of three ways:
SQL_C_TCHAR is declared in a function as follows:
SQLBindParameter(StatementHandle, 1, SQL_PARAM_INPUT, SQL_C_TCHAR, SQL_WCHAR, NameLen, 0, Name, 0, &Name)
When the application is compiled as a Unicode application, the ValueType argument would be changed from SQL_C_TCHAR to SQL_C_WCHAR. When the application is compiled as an ANSI application, the ValueType argument would be changed to SQL_C_CHAR.
Unicode drivers must still support ANSI data types, including SQL_CHAR. If an application working with a non-Unicode driver binds to SQL_WCHAR, the Driver Manager will map the SQL_WCHAR data to SQL_CHAR. However, if an application working with a Unicode driver binds to SQL_CHAR, the Driver Manager will not map the SQL_CHAR data to SQL_WCHAR. The Unicode driver must accept the SQL_CHAR data.
The Driver Manager stores driver and DSN names in Unicode and maps them to ANSI as needed. If a Unicode character cannot be mapped to an ANSI character (as can occur if characters from a code page that is not the native code page of the computer are used in driver and DSN names), the characters that could not be converted are represented by a default character supplied by the system.