The following is a new section added to this appendix.
A Unicode ODBC application sends and retrieves character data primarily in UCS-2. It does this by calling Unicode versions of the ODBC functions (those with a 'W' suffix) and by indicating Unicode data types. The application does not explicitly specify a local code page. The application can still call the ANSI functions and pass local code page strings.
For example, the application may call SQLConnectW() and pass the DSN, User ID and Password as Unicode arguments. It may then call SQLExecDirectW() and pass in a Unicode SQL statement string, and then bind a combination of ANSI local code page buffers (SQL_C_CHAR) and Unicode buffers (SQL_C_WCHAR). The database data types may be local code page or UCS-2 and UTF-8.
If a CLI application calls SQLConnectW or calls SQLSetConnectAttr with SQL_ATTR_ANSI_APP set to SQL_AA_FALSE, the application is considered a Unicode application. This means all CHAR data is sent and received from the database in UTF-8 format. The application can then fetch CHAR data into SQL_C_CHAR buffers in local code page (with possible data loss), or into SQL_C_WCHAR buffers in UCS-2 without any data loss.
If the application does not do either of the two calls above, CHAR data is converted to the applications local code page at the server. This means CHAR data fetched into SQL_C_WCHAR may suffer data loss.
If the DB2CODEPAGE instance variable is set (using db2set) to code page 1208 (UTF-8), the application will receive all CHAR data in UTF-8 since this is now the local code page. The application must also ensure that all CHAR input data is also in UTF-8. ODBC also assumes that all SQL_C_WCHAR data is in the native endian format. CLI will perform any required byte-reversal for SQL_C_WCHAR.
This release of DB2 Universal Database contains the SQLConnectW() API. A Unicode driver must export SQLConnectW in order to be recognized as a Unicode driver by the driver manager. It is important to note that many ODBC applications (such as Microsoft Access and Visual Basic) call SQLConnectW(). In previous releases of DB2 Universal Database, DB2 CLI has not supported this API, and thus was not recognized as a Unicode driver by the ODBC driver manager. This caused the ODBC driver manager to convert all Unicode data to the application's local code page. With the added support of the SQLConnectW() function, these applications will now connect as Unicode applications and DB2 CLI will take care of all required data conversion.
DB2 CLI now accepts Unicode APIs (with a suffix of "W") and regular ANSI APIs. ODBC defines a set of functions with a suffix of "A", but the driver manager does not pass ANSI functions with the "A" suffix to the driver. Instead, it converts these functions to ANSI function calls without the suffix, and then passes them to the driver.
An ODBC application that calls the SQLConnectW() API is considered a Unicode application. Since the ODBC driver manager will always call the SQLConnectW() API regardless of what version the application called, ODBC introduced the SQL_ATTR_ANSI_APP connect attribute to notify the driver if the application should be considered an ANSI or UNICODE application. If SQL_ATTR_ANSI_APP is not set to SQL_AA_FALSE, DB2 CLI converts all Unicode data to the local code page before sending it to the server.